Estimating urban traffic and congestion cost trends for ... · of the social costs arising from...

Estimating urban trafficand congestion cost trendsfor Australian cities

Working PaperNo 71

Bureau of Transport and Regional Economics

Estimating urban traffic and congestion cost trends for Australian cities

Working Paper 71

Bureau of Transport and Regional Economics Department of Transport and Regional Services

Canberra, Australia

© Commonwealth of Australia 2007

ISSN 1440-9707

ISBN 1-921260-10-6

APR2007/50200

This publication is available in hard copy or PDF format from the Bureau of Transport and Regional Economics website at www.btre.gov.au—if you require part or all of this publication in a different format, please contact BTRE.

An appropriate citation for this report is:Bureau of Transport and Regional Economics [BTRE], 2007, Estimating urban traffic and congestion cost trends for Australian cities, Working paper 71, BTRE, Canberra ACT.

Indemnity statementThe Bureau of Transport and Regional Economics has taken due care in preparing the analyses contained in this report. However, noting that data used for the analyses have been provided by third parties, the Commonwealth gives no warranty to the accuracy, reliability, fitness for purpose, or otherwise of the information.

Published byBureau of Transport and Regional EconomicsGPO Box 501, Canberra ACT 2601, AustraliaTelephone (international) +61 2 6274 7210Fax +61 2 6274 6816E-mail: [email protected]: http://www.btre.gov.au

Desktop publishing by Melinda Keane, BTRE. Cover design by Kerry Rose, BTRE.

Printed by Elect printing

Typeset in Optima LT Std and Gill Sans MT [Mac].

iii

Foreword

This report presents the results of a Bureau of Transport and Regional Economics (BTRE) study to identify long-term trends in urban traffic growth, to estimate the consequent impacts of that traffic growth on urban congestion levels, and to attempt a suitable quantification of the social costs arising from those congestion levels. The study deals with the eight Australian capital cities, and presents base case (or business as usual) projections to 2020 of avoidable social costs of congestion for Australian metropolitan traffic. This work updates and revises previous congestion cost projections published by the Bureau (such as Information Sheet 14, BTE 1999), and has been completed to inform the Urban Congestion Review, which was commissioned by the Council of Australian Governments (COAG).

The traffic forecasts contained in this report are derived from BTRE base case projections of Australian transport activity (for the most recent published data see Greenhouse Gas Emissions from Australian Transport: Base Case Projections to 2020, BTRE 2006a). The costing methodology used for this study relies on the results of previous Bureau work using network models to estimate congestion impacts for Australian cities (such as Report 92, Traffic Congestion and Road User Charges in Australian Capital Cities, BTCE 1996b). It is important to note that the current BTRE approach to estimating congestion costs is an aggregate modelling one, i.e., it does not directly use detailed network modelling. Network models generally attempt to simulate the traffic flows on a city’s road system in considerable detail; whereas the aggregate method aims to provide broad estimates of the scale of a city’s congestion situation using aggregate indicators of a city’s overall average traffic conditions.

The main advantage of this aggregate approach relates to the ability to generate congestion cost estimates and projections with much lighter computational and information resources than required by data-intensive network or microsimulation models. The main disadvantage relates to the approximate nature of such aggregate costings—with congestion being such a non-linear, inhomogeneous and stochastic

process, highly accurate, location-specific assessments of its impacts can typically only be accomplished using detailed network models.

Aggregate national congestion cost estimates do not serve in any way to replace the results of detailed traffic simulation models conducted at a jurisdictional level (for both existing or future studies)—but are intended as a complement to their more in-depth findings. The results presented in this report are therefore provided as ‘order of magnitude’ evaluations—to help with considerations dealing with the likely aggregate costs of urban transport externalities for Australia, and their likely future trends. However, even with these caveats, the forecasts suggest an appreciable increase in the social costs of congestion over the next 15 years under a business-as-usual scenario.

The BTRE acknowledges the important contributions made by State and Territory colleagues through the Urban Congestion Management Working Group of the Standing Committee on Transport (SCOT). The study was undertaken by Dr David Cosgrove and Dr David Gargett.

Phil Potterton Executive Director Bureau of Transport and Regional Economics April 2007

iv

BTRE | Working Paper 71

Contents

Foreword ...................................................................................................... iii

At a glance ................................................................................................... xv

Summary ........................................................................................................ 1

Background ............................................................................1

Report outline........................................................................3

Methodology .........................................................................7

Congestion cost estimates ................................................13

Part 1 – Traffic growth ...............................................................................21

Growth in urban passenger demand ..............................21

Urban freight growth .........................................................40

Car traffic growth ................................................................43

Bus and motorcycle traffic growth ..................................47

Truck traffic growth ............................................................48

Projections of total traffic for Australian cities ..............50

Part 2 – Social costs of congestion .........................................................55

Increasing traffic trends underlying congestion ..........55

The costs of congestion .....................................................77

Total congestion cost estimation .....................................84

Total delay cost values ........................................................92

Congestion estimation method—avoidable social costs ...........................................................................99

Avoidable social cost values ........................................... 107

Projections of congestion costs ......................................107

Uncertainty and sensitivity of estimation process ......113

Current data deficiencies and possible future directions ............................................................................121

Appendix – Aggregate inputs and supplementary results ...............127

References .................................................................................................135

Further readings .......................................................................................139

v

vi

Figures

Figure S.1 Average unit costs of congestion for Australian metropolitan areas—current and projected ....................................................................16

Figure S.2 Trends in avoidable social costs of congestion—scenario results .................................20

Figure 1.1 Historical and projected urban passenger movement, Australian metropolitan total ............23

Figure 1.2 Historical and projected modal share, Australian metropolitan passenger travel ............23

Figure 1.3a Historical trend in metropolitan passenger travel ........................................................26

Figure 1.3b Historical trend in passenger mode share ...........27

Figure 1.3c Relationship of per capita travel to per capita income ............................................................27

Figure 1.4a Historical trend in Sydney passenger travel ........28

Figure 1.4b Historical trend in Sydney passenger mode share ................................................................28

Figure 1.4c Relationship of per capita travel to per capita income, Sydney .........................................................29

Figure 1.5a Historical trend in Melbourne passenger travel ............................................................................29

Figure 1.5b Historical trend in Melbourne passenger mode share ................................................................30

Figure 1.5c Relationship of per capita travel to per capita income, Melbourne .................................................30

Figure 1.6a Historical trend in Brisbane passenger travel .....31

vii

Figure 1.6b Historical trend in Brisbane passenger mode share ................................................................31

Figure 1.6c Relationship of per capita travel to per capita income, Brisbane ......................................................32

Figure 1.7a Historical trend in Adelaide passenger travel .....32

Figure 1.7b Historical trend in Adelaide passenger mode share ............................................................................33

Figure 1.7c Relationship of per capita travel to per capita income, Adelaide ......................................................33

Figure 1.8a Historical trend in Perth passenger travel ...........34

Figure 1.8b Historical trend in Perth passenger mode share ............................................................................34

Figure 1.8c Relationship of per capita travel to per capita income, Perth ............................................................35

Figure 1.9a Historical trend in Hobart passenger travel ........35

Figure 1.9b Historical trend in Hobart passenger mode share ............................................................................36

Figure 1.9c Relationship of per capita travel to per capita income, Hobart .........................................................36

Figure 1.10a Historical trend in Darwin passenger travel ........37

Figure 1.10b Historical trend in Darwin passenger mode share ............................................................................37

Figure 1.10c Relationship of per capita travel to per capita income, Darwin .........................................................38

Figure 1.11a Historical trend in Canberra passenger travel ....38

Figure 1.11b Historical trend in Canberra passenger mode share ............................................................................39

Figure 1.11c Relationship of per capita travel to per capita income, Canberra .....................................................39

Figure 1.12 Historical and model fitted data for aggregate freight task .................................................................42

viii

Figure 1.13 Cross section, time-series data and model fit for city freight tasks .............................................43

Figure 1.14 Historical trend in annual per capita passenger vehicle travel versus real Australian income per capita .................................................................... 44

Figure 1.15 Historical trend in components of Australian population increase .................................................46

Figure 1.16 Projected travel by motor vehicles, Australian metropolitan total .....................................................53

Figure 1.17 Total projected traffic for Australian capital cities ............................................................................53

Figure 2.1 Historical and projected urban passenger movement, Australian metropolitan total ............55

Figure 2.2 Historical and projected modal share, Australian metropolitan passenger travel ............57

Figure 2.3 Historical and projected travel by motor vehicles, Australian metropolitan total .................58

Figure 2.4 Total projected car traffic for Australian capital cities ...............................................................59

Figure 2.5 Total projected LCV traffic for Australian capital cities ...............................................................59

Figure 2.6 Total projected rigid truck traffic for Australian capital cities ...............................................................60

Figure 2.7 Total projected articulated truck traffic for Australian capital cities ............................................60

Figure 2.8 Total projected bus traffic for Australian capital cities ...............................................................61

Figure 2.9 Total projected motorcycle traffic for Australian capital cities ............................................61

Figure 2.10 Total projected traffic for Australian capital cities–VKT ...................................................................62

ix

Figure 2.11 Total projected traffic for Australian capital cities–PCU VKT ..........................................................62

Figure 2.12 Total projected VKT for Sydney ..............................63

Figure 2.13 Total projected VKT for Melbourne ......................63

Figure 2.14 Total projected VKT for Brisbane ...........................64

Figure 2.15 Total projected VKT for Adelaide ...........................64

Figure 2.16 Total projected VKT for Perth .................................65

Figure 2.17 Total projected VKT for Hobart ..............................65

Figure 2.18 Total projected VKT for Darwin .............................66

Figure 2.19 Total projected VKT for Canberra ..........................66

Figure 2.20 Representative functional forms for normalised speed-flow relationships for various road types ....................................................82

Figure 2.21 Typical response curves for average fuel consumption to variation in vehicle speed .........83

Figure 2.22 Typical response curves for petrol car emission rates to variation in speed, by pollutant type .......................................................83

Figure 2.23 Hourly traffic volumes for typical metropolitan travel ...................................................84

Figure 2.24 Typical daily traffic profile for commercial vehicles .......................................................................85

Figure 2.25a Typical daily VKT profile by vehicle type ..............85

Figure 2.25b Typical daily VKT profile by road type .................86

Figure 2.26 Typical daily passenger task profile by vehicle type ................................................................87

Figure 2.27 Average urban traffic performance by time of day .................................................................87

Figure 2.28 Typical delay profile .................................................89

x

Figure 2.29 Typical daily delay profile by vehicle type ...........89

Figure 2.30 Divergence in estimated delay depending on comparison speeds ............................................90

Figure 2.31 Distribution of total delay costs by vehicle type ................................................................92

Figure 2.32 Representative variation of trip reliability with traffic volume ....................................................94

Figure 2.33 Typical daily profile of total time costs .................96

Figure 2.34 Projected daily VKT profile ....................................97

Figure 2.35 Growth in Melbourne freeway volumes ..............98

Figure 2.36 Projected daily profile of total time cost ..............98

Figure 2.37 Basic economic theory of congestion costs ......100

Figure 2.38 Modelled average and marginal cost curves .....103

Figure 2.39 Proportion of external costs for a typical divided arterial ........................................................104

Figure 2.40 Proportion of external costs .................................105

Figure 2.41 Proportion of external costs and deadweight losses ..................................................106

Figure 2.42 Typical daily profile for avoidable costs of congestion ...........................................................107

Figure 2.43 Projected daily cost profile ..................................108

Figure 2.44 Projected avoidable costs of congestion by city ........................................................................112

Figure 2.45 Composition of projected avoidable costs of congestion ................................................113

Figure 2.46 Sensitivity scenarios for major parameter variations ..................................................................117

Figure 2.47 Sensitivity scenarios for variation in the value of time lost ....................................................118

xi

Figure 2.48 Sensitivity scenarios for UPT share variation ....120

Figure 2.49 Sensitivity scenarios for travel demand variation ....................................................................120

Figure A.1 Long-term trends in national per capita passenger and freight movement relative to per capita income levels ...................................129

Figure A.2 Summary flowchart–BTRE base case projections for traffic growth ...............................132

Figure A.3 Summary flowchart–BTRE method for congestion cost estimation ...................................133

Figure A.4 Forecast fuel price trend for BTRE base case demand projections .....................................134

xii

Tables

Table 1.1 Capital city road freight task—trends and projections .........................................................41

Table 1.2 Example car traffic projections for Australian cities .........................................................47

Table 2.1 National base case projections of metropolitan vehicle kilometres travelled by type of vehicle, 1990–2020 ..................................67

Table 2.2 Base case projections of vehicle kilometres travelled by type of vehicle for Sydney, 1990–2020 ...............................................68

Table 2.3 Base case projections of vehicle kilometres travelled by type of vehicle for Melbourne, 1990–2020 ....................................................................69

Table 2.4 Base case projections of vehicle kilometres travelled by type of vehicle for Brisbane, 1990–2020 ....................................................................70

Table 2.5 Base case projections of vehicle kilometres travelled by type of vehicle for Adelaide, 1990–2020 ....................................................................71

Table 2.6 Base case projections of vehicle kilometres travelled by type of vehicle for Perth, 1990–2020 ....................................................................72

Table 2.7 Base case projections of vehicle kilometres travelled by type of vehicle for Hobart, 1990–2020 ....................................................................73

Table 2.8 Base case projections of vehicle kilometres travelled by type of vehicle for Darwin, 1990–2020 ....................................................................74

xiii

Table 2.9 Base case projections of vehicle kilometres travelled by type of vehicle for Canberra, 1990–2020 ....................................................................75

Table 2.10 National base case projections of total metropolitan vehicle travel—passenger car equivalents, 1990–2020 .............................................76

Table 2.11 Base case projected estimates of social costs of congestion for Australian metropolitan areas, 1990–2020 ......................................................109

Table 2.12 Base case projections for average network delay due to congestion for Australian metropolitan areas, 1990–2020 .............................110

Table 2.13 Base case projections for average unit costs of congestion for Australian metropolitan areas, 1990–2020 ......................................................111

Table A.1 State and territory population projections ........127

Table A.2 Capital city population projections .....................128

Table A.3 Base case GDP growth assumptions ...................129

Table A.4 Base case projected estimates of social costs of congestion for business vehicle travel, 1990–2020 ..................................................................130

Table A.5 Base case projected estimates of social costs of congestion for private vehicle travel, 1990–2020 ..................................................................131

xv

At a glance

• This report has been completed to inform the Urban Congestion Review, commissioned in February 2006 by the Council of Australian Governments.

• Total travel in Australian urban areas has grown ten-fold over the last 60 years. Private road vehicles now account for about 90 per cent of the total urban passenger task (up from around 40 per cent in the late 1940s). The current trend of near linear increases in aggregate urban traffic is forecast to continue over the projection period, with total kilometres travelled growing by 37 per cent between 2005 and 2020. Commercial vehicle traffic is forecast to grow substantially more strongly (averaging around 3.5 per cent per annum) than private car traffic (at about 1.7 per cent per annum).

• BTRE estimates of the ‘avoidable’ cost of congestion (i.e. where the benefits to road users of some travel in congested conditions are less than the costs imposed on other road users and the wider community) for the Australian capitals (using an aggregate modelling approach) total approximately $9.4 billion for 2005. This total is comprised of $3.5 billion in private time costs, $3.6 billion in business time costs, $1.2 billion in extra vehicle operating costs, and $1.1 billion in extra air pollution costs. The estimates do not take account of the implementation costs of any congestion alleviation measures, and are not strictly comparable to standard measures of aggregate national income (such as GDP).

• BTRE base case projections have these social costs of congestion rising strongly, to an estimated $20.4 billion by 2020. The city specific levels rise from $3.5 billion (2005) to $7.8 billion (2020) for Sydney, $3.0 billion to $6.1 billion for Melbourne, $1.2 billion to $3.0 billion for Brisbane, $0.9 billion to $2.1 billion for Perth, $0.6 billion to $1.1 billion for Adelaide, $0.11 billion to $0.2 billion for Canberra, about $50 million to $70 million for Hobart, and $18 million to $35 million for Darwin.

• The complex nature of congestion effects leads to reasonable levels of uncertainty in such cost estimations. However, irrespective of questions over exact dollar valuations of congestion costs, sensitivity testing implies that, in the absence of improved congestion management, it will be challenging to avoid escalating urban congestion impacts, given the rising traffic volumes expected within the Australian capital cities.

Summary

Background

In February 2006, the Council of Australian Governments (COAG) commissioned a review of urban congestion trends, impacts and solutions (COAG 2006).

The Bureau of Transport and Regional Economics (BTRE) was subsequently commissioned to undertake a study to:

1. Examine main current and emerging causes, trends and impacts of urban traffic growth and congestion.

2. Identify any deficiencies in information and make recommendations regarding the collection and sharing of nationally consistent data.

The BTRE study (which commenced prior to the COAG Review, as part of the BTRE research program) analyses a number of the underlying factors contributing to the growth in traffic volumes and congestion intensity (across the metropolitan road networks), and the impact on traffic delays, travel reliability, and congestion’s spread in both area and duration. The study then provides estimates of the economic costs of these growing time losses, and other social and economic consequences of growth in urban traffic congestion.

The study estimates the level of average trip delay (and other social impacts) of current congestion levels and forecasts how these will likely vary over time. A consistent aggregate methodology for estimating the average avoidable social costs of congestion is presented, along with projected future cost levels (which have implications for identifying the most economically-efficient level of congestion management).

It is important to stress that the current BTRE approach to estimating congestion costs is an aggregate modelling one, i.e., it does not

1

use detailed network models. Network models generally simulate the traffic flows on all of a city’s major roads, often in great detail; whereas our aggregate method aims to roughly estimate the scale of a city’s congestion situation using aggregate indicators of a city’s overall average traffic conditions. The main advantages of this aggregate approach relate to the ability to generate congestion cost estimates and projections with relatively slight computational and data resources. (Network or microsimulation models tend to require extensive data input and a considerable level of computational and maintenance support.) The main disadvantages relate to congestion being such a non-linear, inhomogeneous and stochastic process that highly accurate assessments of its impacts can really only be accomplished using detailed network models. As yet, however, there are no complete estimates of the cost of congestion (for Australian cities) using a network modelling approach.

The congestion cost estimates presented in this study are therefore provided as ‘order of magnitude’ evaluations—to help with considerations dealing with the likely aggregate costs of urban transport externalities for Australia, and their likely future trends. Detailed assessments of each of the cities’ congestion impacts, especially analyses at the level of particular major city roads or thoroughfares, should continue to be pursued at the jurisdictional level using the appropriate network modelling frameworks. In fact, the BTRE is hopeful that not only will the various jurisdictions continue to develop traffic and congestion modelling techniques on their road networks, but that the results of such modelling will continue to improve understanding of congestion occurrence in Australian cities (and lead to more consistent comparisons being made between studies done in the various cities).

Though the congestion costing methodology used for this study does not directly rely on running detailed network models, the approach has been based on the results of network modelling of Australian cities’ road systems—both from the literature and from previous Bureau work using network models to estimate congestion impacts (for example, Report 92, Traffic Congestion and Road User Charges in Australian Capital Cities, BTCE 1996b). The modelling of congestion for this study also draws on other previous Bureau work, including the results of BTCE (1996a) Report 94 (see Chapter 18 and Appendix XIII), Information Sheet 16, BTE (2000) and Urban Pollutant Emissions from Motor Vehicles: Australian Trends to 2020 (BTRE 2003a).

2


The traffic forecasts contained in this report are derived from base case (or ‘business as usual’) projections of Australian transport sector activity. Projections are provided to 2020. This work updates previous BTRE projections of aggregate urban transport tasks using a suite of different demand models—utilising various modelling approaches (e.g. structural, econometric and dynamic fleet models) to project the different modal task components, and by aggregating estimates of demand levels for various transport sub-sectors to obtain sector totals (often termed a ‘bottom-up’ modelling approach). Previous Bureau reports published on our aggregate demand modelling include BTRE (2002a) Report 107, BTCE (1995) Report 88, BTCE (1996a) Report 94, BTCE (1997) Working Paper 35 and BTRE (2006c) Report 112. Extensions and revisions to Report 107 have also been prepared in more recent years (BTRE 2003a, 2003b, 2006b).

Report outline

This BTRE Working Paper is composed of two sections:

• Part I—dealing with trends in transport tasks (urban passenger and freight movement) and traffic growth for the capital cities; and

• Part 2—dealing with the estimation of the costs of congestion for the capital city road networks.

The report gives updated forecasts of urban transport demand and updated forecasts of the consequent traffic growth in Australian cities. The forecasting process entails examining the factors influencing the demand for urban road use, and assessment of trends in population growth and distribution, income growth, and demand for urban freight, service provision and passenger movement. The transport task estimates address trends in public transport patronage and the implications for forecasts of total vehicle kilometres travelled (VKT).

Part 2 of the report provides updated estimates of congestion trends and the likely magnitude of the social costs of congestion associated with the traffic growth patterns described in Part 1. The methodology assesses the impacts that changing travel volumes are likely to have on changing average traffic profiles and the distribution of vehicle

3

Summary

traffic and congestion. The models contain a number of adjustable input parameters and default city-by-city calibrations that can be updated or revised if more detailed or network-specific data later become available.

Due to the complexity of the congestion problem, and the difficulty in collating enough information to model every intricate facet of road system performance, the modelling has had to include a variety of simplifying assumptions and approximations. Some of the major areas of modelling uncertainty surround the issues of:

• Possible peak-spreading of traffic volumes in the future (especially considering some major urban freeways are already operating at close to full capacity during the peak hours). The BTRE model allows for a degree of future peak-spreading (the effects on the estimates, if actual future spreading is greater or smaller than currently modelled, is addressed in a latter section of Part 2 on model sensitivities).

• What exact value do different parts of the community place on their time (for example, whether people value waiting time more highly than trip time reliability; whether delays during various trip purposes are felt more strongly than during others; or whether delay during longer trips is worse than during shorter trips). Basically, two of the most important uncertainties concern the question of the most suitable dollar value for an hour of time lost to congestion delay (either for business road use or for private travel), and the question of what proportions of urban trips are not highly time-sensitive. (The dependence of the estimates on the default parameter settings for average values of time is also addressed in Part 2).

• Comparison speed values. When calculating how much delay congestion causes, recorded average traffic speeds have to be compared with a less-congested benchmark speed (such as the free-flow speed—how fast a car is able to travel at, on that particular road link, if no other vehicles are present). For actual road systems, the definition and estimation of such speeds is typically complicated and fairly approximate (often involving limited floating-car surveys).

4


• Supply-side factors for urban road networks (such as, likely trends in future network capacity expansion—including both further road construction and other transport infrastructure development; and medium to longer term plans for urban development and provision of transport services). The BTRE base case scenario for congestion cost projections assumes that an increase of about one per cent per annum (in total available lane-kilometres) would be representative of long-term road provision trends for our capital cities, and continues this trend to 2020. This assumption could prove to be conservative, since traffic management systems within our cities will also tend to be developed further over time, and could also serve to increase overall future network capacity. However, addressing the likely impacts of such progression in travel management practices is probably best handled within possible further work on congestion reduction scenarios—rather than the business-as-usual projections of this study.

• Demographic effects—which can influence not only total demand levels but also elements of trip distribution (for example, peak-spreading of trips may increase in the future at even faster rates, as the average age of the population and retirement rates increase, and a growing proportion of travellers are not constrained to travel during the standard work-day peak hour). Though the BTRE demand projections make allowances for a fairly wide range of demographic effects (e.g. trends in average age of the population, average employment levels and typical household sizes), these analyses—like the current BTRE traffic flow modelling—are also at an aggregate level. Modelling approaches dealing with specific future changes to urban form and with a fine-grained level of demographic detail (i.e. dealing with each city’s population and employment distribution) are beyond the scope of this study.

• The possible flow-on effects of urban congestion. The costs presented in this report primarily relate to estimated valuations of excess travel time for road users (with some allowance for the inefficiency of vehicle engine operation under stop-start conditions, leading to higher rates of fuel consumption and pollutant emissions). However, there is a series of other possible consequences of urban congestion—ranging from some businesses having to re-locate or close (due to restrictions on their operations from congestion delays), to widespread psychological

5

Summary

stress and irritation from coping with heavy traffic levels, to reducing the efficiency of public transit and the attractiveness of transit or non-motorised transport options. Estimation of costs for these wider effects is difficult (with the literature covering suitable quantification methods being, so far, very limited), and also tends to be beyond the scope of this study (though this issue is briefly considered in the sensitivities section of Part 2).

Related to this final dot point, is the issue of how many other costs, besides the average delay experienced on travel in heavy traffic, are already implicitly borne by a city’s population due to the level and distribution of current urban congestion. For example, land and housing values are typically much higher in areas (such as those in the centre of the city) that allow shorter commuting times and better transport access to major amenities. There are also issues of wasted infrastructure resources where unmanaged congestion on some urban freeways causes flow breakdowns that reduce vehicle throughputs, to only a small proportion of the road’s design capability, for much of the peak periods. As well, there are typically costs associated with urban decentralisation—which is often prompted by rising congestion levels—where the prices for a wide variety of urban goods and services will partially reflect the difficulty in delivering such goods and services over a widespread, densely trafficked network.

Even though congestion is to a certain extent self-limiting (that is, if enough people try to enter an already busy traffic stream, then the congestion will eventually get so severe that some are deterred from travelling), there are still costs associated with such feedback constraints. Not only will average queuing times and flow disruptions continue to increase for those choosing to still travel, but those having their travel deferred (to a less convenient time) or discouraged by congestion suffer a loss of utility. Such losses in consumer utility mean that even natural behavioural effects that tend to improve the incidence of peak congestion levels (such as trip timing decisions resulting in peak-spreading) will also have some associated costs to travellers. Thus the broad pricing question could be framed in terms of whether better alternatives exist than to rely on the costs implicit in such individual and community ‘self regulation’. Estimates of the aggregate social costs of congestion, such as those provided by this report, help to focus on the search for the most cost-effective solutions.

6


Part 2 of the report closes with a brief summary of the BTRE’s assessment of the main data shortcomings and information gaps (within the current dataset collection processes) that have a bearing on congestion intensity and traffic demand management. Basically, various jurisdictions’ on-going Traffic Systems Performance Monitoring tends to collect a reasonable amount of useful congestion-related data (such as details on average traffic flows), but there still appears to be a relative scarcity of quality data on:

• the details of urban freight distribution (and therefore on the impacts of congestion on freight costs);

• variability in average trip times (where road users appear to place a high premium on travel time reliability); and

• the composition of the traffic mix and how it varies over the hours of the day (not only in terms of vehicle type, but also with regards to the purposes or destinations of the various journeys).

As greater quantities of data continue to be collected in real-time, there would undoubtedly be social benefits in the greater provision of real-time traffic information systems, e.g, internet sites showing current traffic performance statistics to aid travellers’ trip planning, allowing trips to be rescheduled or re-routed (if notified of a particular incident or irregular delay on their usual route).

Methodology

The BTRE aggregate approach is summarised below. (Also, for a schematic diagram roughly summarising the main steps involved in the estimation and projection process, see Appendix figures A.2 and A.3.)

Note that the current methodology addresses estimated traffic volumes and congestion costs for metropolitan areas—i.e. the Statistical Divisions of the eight State/Territory capital cities. The estimates would be higher to some extent if all the regional urban areas, such as Newcastle, Geelong or parts of South-East Queensland, were also included.

7

Summary

Passenger traffic forecast Trends in passenger travel in each capital reveal that:

(1) Total passenger kilometres per person have been increasing as income per person increase.

(2) However, for most cities that increase in travel per person has just about saturated.

(3) Therefore, the rate of increase in total city travel will tend, in the near future, to fall and equal the rate of population growth of the city concerned.

(4) Furthermore, in almost all cities, the share of car and urban public transport has been virtually constant for around 30 years (typically about 90 per cent of motorised passenger-kilometres done in light motor vehicles and about 10 per cent for rail, bus and ferry).

Therefore, at a base level, forecasting of car travel can be done (within a business-as-usual scenario) by (1) relating car travel per person to income per person and (2) multiplying resulting car-kilometres per capita (derived from projected income levels) by projected population levels.

Freight traffic forecast Trends in freight haulage in each capital have been related to (national) GDP per person. Estimated freight task per capita are then also multiplied by the city population. Assumptions are then made about the likely split between vehicle types (light commercial vehicles, rigid trucks and articulated trucks) and about future average loadings per vehicle (by vehicle type).

The result is a forecast of aggregate vehicle kilometres by vehicle type for each city.

Resulting traffic forecasts—freight and passenger vehiclesThe forecast growth in car traffic in the cities tends to decelerate over time, in the base case, due to the modelled saturating trend of car travel per person against income.

8


Balancing this slowing car growth is rapid growth in the light commercial vehicle (LCV) fraction of the traffic (which is already a substantial part of the traffic stream). Heavy trucks also grow quickly but from the base of a small fraction of the current traffic stream. Annual growth in total VKT by LCVs has averaged between 3 and 4 per cent for well over 20 years, and the base case essentially continues this trend to 2020, with continued (projected) economic growth leading to continued VKT growth. This relatively high level of commercial traffic growth is predicated on the assumption that there will be no decoupling of activity in the freight and service sectors from overall income trends (i.e. GDP per person) during the projection period. When (or if) such decoupling does occur, as has already happened for the passenger sector, VKT growth for LCVs and trucks will also be expected to decelerate, but the evidence suggests that such a saturating trend in per capita freight movement is unlikely in the short to medium term (e.g. see Appendix figure A.1).

The result is near linear increases in traffic over the time period investigated for recent congestion trends (i.e. between 1990 and the present), continuing over the projection period (i.e. the present to 2020). In other words, approximately as much traffic in absolute terms will be added to the average city network in the next 15 years as was added in the past 15 years.

For aggregate metropolitan traffic growth, the BTRE base case projections have total annual kilometres travelled (in passenger car equivalent units, PCU-km) increasing by close to 37 per cent between 2005 and 2020—from around 138 billion PCU-km in 2005 (across all eight capital cities) to 188 billion in 2020. By city, the projected PCU-km increases (using base case input assumptions to the BTRE models) are about 38 per cent for Sydney, 33 per cent for Melbourne, 46 per cent for Brisbane, 27 per cent for Adelaide, 44 per cent for Perth, 13 per cent for Hobart, 40 per cent for Darwin and 29 per cent for Canberra. Variations in forecast city growth rates for VKT (e.g. the high growth in Brisbane, Perth and Darwin, and the low growth in Hobart) are due largely to variations in projected population growth.

The costs of congestion Congestion imposes significant social costs with interruptions to traffic flow lengthening average journey times, making trip travel

9

Summary

times more variable and making vehicle engine operation less efficient. The (generalised) cost estimates presented in this report include allowances for:

• extra travel time (e.g. above that for a vehicle travelling under less congested conditions),

• extra travel time variability (where congestion can result in trip times becoming more uncertain—leading to travellers having to allow for an even greater amount of travel time than the average journey time, in order to avoid being late at their destination),

• increased vehicle operating costs (primarily higher rates of fuel consumption), and

• poorer air quality (with vehicles under congested conditions emitting higher rates of noxious pollutants than under more freely flowing conditions, leading to even higher health costs).

Three different ‘cost of time delay’ calculations are often made, specifically:

(1) Total Cost of Congestion Estimate—total delay values use actual travel speeds versus estimated free flow speeds (e.g. versus what speed one could typically average travelling across the city in the middle of the night).

(2) External Cost of Congestion Estimate—still based on actual travel speeds versus free flow speeds, but estimates that portion of total costs that road users impose on others (through not having to personally meet the total costs caused by their travel decisions).

(3) Deadweight Loss Cost of Congestion—the increase in net social benefit if appropriate traffic management or pricing schemes were introduced and optimal traffic levels were obtained. Basically, the problem with a currently congested traffic flow is that it includes a quantity of travel for which the total costs to society exceed the benefits of that element of travel. The net loss on this amount of travel (after converting hours lost to delays into dollar terms) is given by the deadweight loss (DWL). Avoiding this loss would produce a net social benefit—leading to the common descriptions of such DWL values as a measure of the ‘cost of doing nothing about congestion’, or the ‘avoidable cost of traffic congestion’.

10


Though somewhat more involved to estimate, the third cost definition is generally preferable from a theoretical standpoint—especially for policy assessment. The ‘avoidable cost of congestion’ is a direct measure of what can, in principle, actually be achieved by tackling the congestion problem. It is typically of much greater policy relevance than total delay costs, because there is a possibility of obtaining the net benefits described in the third definition (i.e. given by DWL valuations), while it will not typically be feasible to reduce total delay costs to zero for real-world traffic streams.

Therefore, the primary values derived by this study to refer to the ‘social costs of congestion’ are the estimated deadweight losses (DWLs) associated with a particular congestion level—which, reiterating, give a measure of the costs of doing nothing about congestion or the avoidable costs of traffic congestion. That is, DWL valuations give an estimate of how much total costs (for time lost and other wasted resources) could be reduced if traffic volumes were reduced to the economically optimal level. This optimal level is defined as the traffic volume (and distribution) that would result if, for a given travel demand, the generalised cost that motorists based their trip decisions on was equal to the marginal travel cost rather than on their private, individual travel costs (i.e. on the current average generalised travel cost). That is, if through an appropriate transport demand management or pricing instrument, each motorist choosing to enter already congested traffic had to take account of not only their personal travel time costs, but also the cost of all the extra delay that their entry into the traffic stream is likely to impose on others.

Essentially, there is no way to avoid all the total delay costs since actual traffic volumes cannot travel at free flow speeds at all times. Whereas the portion of total costs that could theoretically be saved, if some traffic management strategy was capable of changing traffic conditions to the economically optimal level, can be estimated by the deadweight loss associated with that change in traffic level. A DWL value will still tend to be an upper bound for the actual social benefits achieved by any particular congestion reduction strategy since it would take a perfectly variable management scheme, that targeted congestion by exact location and time of day (depending on the changing traffic levels on each of the network’s road links), to obtain the economic optimum. However, it will be a substantially closer guide to actual obtainable benefits than a total delay cost estimate.

11

Summary

Of course, actually putting a particular traffic management strategy into practice will also tend to involve implementation, possible extra infrastructure, and on-going administration costs—all of which will have a bearing on how much of the full theoretical benefit level would in fact be gained. That is, the (avoidable social) cost values given in this report do not directly refer to actual obtainable savings for congestion reduction measures since the introduction and running costs will vary from measure to measure (and in some cases will be considerable), and are not taken account of within the congestion cost valuations of this report.

A related point is that even though the congestion costs derived by this study are considerable (as is typically the case for any such derivations that use standard assumed values of time), the dollar values obtained are not directly comparable to aggregate income measures (such as GDP). Some elements involved in the costings would have GDP implications (e.g. the timeliness and reliability of freight and service deliveries will impact on business productivity levels). However, a major proportion of the derived cost values refer to elements that play no part in the evaluation of GDP, such as private travel costs. Say, for example, that some congestion measure did happen to successfully reduce a city’s traffic to the economic optimum, thereby ‘saving’ the avoidable congestion costs, and resulting in many travellers being able to take their trips less encumbered by congestion delays. Though these time savings would undoubtedly have benefits for many road users (where the DWL calculations attempt to suitably evaluate the net changes in road user utility in dollar terms), such benefit amounts—especially with regard to private individuals—will not have a clear-cut bearing on the size of any GDP changes that happen to flow from the congestion reduction.

When calculating the avoidable delay costs, the BTRE congestion model separately estimates travel characteristics by private cars, business cars, LCVs, buses and trucks, and uses different hourly delay cost rates for each category. The models use speed-flow curves, for various road types and city areas, to derive aggregate traffic delay estimates by time of day.

Various adjustable parameters are used in the modelling, including a cost for variability of travel time, percentages of short trips (where delay will not be noticeable), and an allowance for a proportion of

12


total trips to be less time-sensitive than average. The model also allows for increasing delay costs in peak traffic periods (which is capable of generating some peak spreading).

Once time delay has been calculated, and converted into dollar terms (using appropriate values of time), other social costs associated with congestion (i.e. extra vehicle operating costs and air pollution costs) are also calculated. These other costs are added to the estimates of deadweight losses (associated with the calculated level of aggregate delay) to generate our national estimates for the Avoidable Social Costs of Congestion.

Congestion cost estimates

BTRE aggregate congestion estimates for this study give a total of about $9.4 billion for the 2005 social costs of congestion1 (on the basis of potentially avoidable costs, calculated from the deadweight losses associated with current congestion levels across the Australian capitals). This total is comprised of approximately $3.5 billion in private time costs (losses from trip delay and travel time variability), $3.6 billion in business time costs (trip delay plus variability), $1.2 billion in extra vehicle operating costs, and $1.1 billion in extra air pollution damage costs. The national total is spread over the capital cities, with Sydney the highest (at about $3.5 billion), followed by Melbourne (with about $3.0 billion), Brisbane ($1.2 billion), Perth ($0.9 billion), Adelaide ($0.6 billion), Canberra ($0.11 billion), Hobart ($50 million) and Darwin ($18 million).

BTRE aggregate projections (using the base case scenario for future traffic volumes) have the avoidable social costs of congestion more than doubling over the 15 years between 2005 and 2020, to an estimated $20.4 billion. Of this $20.4 billion total, private travel is forecast to incur time costs of approximately $7.4 billion (DWL of trip delay plus trip time variability), and business vehicle use $9 billion (DWL of trip delay plus variability). Extra vehicle operating costs contribute a further $2.4 billion and extra air pollution damages a further $1.5 billion. The city specific levels rise to approximately $7.8 billion for Sydney, $6.1 billion

13

Summary

1 In estimated costs to Australian economic welfare, and not in terms of measured economic activity or gross domestic product (and therefore not directly referable to as a proportion of GDP).

for Melbourne, $3.0 billion for Brisbane, $1.1 billion for Adelaide, $2.1 billion for Perth, $0.07 billion for Hobart, $35 million for Darwin, and $0.2 billion for Canberra.

As mentioned previously, it should be stressed that even though these cost levels are theoretically avoidable, they do not directly relate to any net savings that may be possible under any particular congestion abatement policies, and do not provide an explicit (or directly quantifiable) estimate for the magnitude of any changes to GDP resulting from such policies. A DWL valuation gives an (order-of-magnitude) estimate of the worth society places on the disadvantages of current congestion related delays and transport inefficiencies, relative to travel under less-dense traffic conditions (i.e. at economically optimal traffic levels). The (DWL-derived) aggregate cost values given in this report:

• do not allow for any of the wide range of costs that would be associated with actually implementing specific traffic management measures—where the introduction of any measure aimed at congestion reduction will typically incur both set-up and ongoing operating costs (which will have to be considered separately from any of the benefits arising from changes in consumer utility that are estimated by DWL reductions); and

• do not evaluate how the flow-on effects of congestion reductions will impact on general economic activity. While personal travel costs influence the availability and cost of labour, the largest component of these estimated costs (i.e. travel time) is not directly measured in GDP. Though congestion reductions will typically have productivity benefits (e.g. through reductions in labour and housing costs), the difficult question of precisely quantifying all the flow-on impacts on measured GDP is beyond the scope of this study.

Note that, even after allowing for the issue of implementation costs, many studies that give congestion cost estimates are less than fully suitable for assessing the actual impacts of congestion on society—by which we mean suitably valuing the avoidable social costs of congestion (i.e. those costs that potentially could be saved under appropriate policy interventions). This is because such studies are usually based on the total time costs for congestion delay which, as discussed previously, refer to the differences in costs between average

14


travel at congested traffic levels and travel under totally uncongested or free-flow conditions. Total costs of congestion delay (i.e. ‘the costs of congestion relative to free-flow’) estimates are derived by the BTRE methodology but only as intermediate values, used to then derive the DWL values (‘the cost of doing nothing about congestion’).

Since total delay values are based on the value of the excess travel time compared with travel under completely free-flow conditions—an unrealisable situation for actual road networks—they are rather poor measures of the social gains that could be obtained through actual congestion reduction, and are typically only given in this paper for comparison purposes.

For example, to compare with those studies quoting only total delay costs, the current BTRE analysis has an Australian total of $11.1 billion for total annual delay costs over the eight capitals for 2005 (that is, excluding the cost elements for trip variability, vehicle operating cost and air pollution)—compared with around $5.6 billion for the preferable deadweight loss valuation of delay.

The base case demand projections have the value of total metropolitan delay (i.e. cost of excess travel time compared with completely free-flow conditions, and excluding the cost elements for trip variability, vehicle operating cost and air pollution) rising to about $23 billion by 2020—compared with the portion of this total cost of about $12.6 billion for the (DWL-derived) avoidable costs of traffic congestion delay for 2020.

Note that the estimated levels of total delay costs given here are lower than values previously issued in BTCE Information Sheet 14 for total delay costs, yet allowing for methodological differences, the national results are of a similar order of magnitude. In fact, if allowance is made for the parameters in the current methodology that tend to lower the cost estimates, relative to the much more simplistic derivations in Information Sheet 14—such as, for a percentage of trips to be below the threshold of noticeable delay, for a percentage of trips to be less time sensitive than average, for vehicle occupancy to vary over the day, for lower delay levels experienced on local roads, and for travel to peak-spread where possible—then the national total delay cost levels are quite comparable. However, the city-by-city composition of the updated congestion estimates presented here (see figure S.1) are considerably different from the previous Bureau values given in Information Sheet 14.

15

Summary

The average unit costs of congestion (that is, total avoidable congestion costs for metropolitan Australia divided by total VKT in PCU-km terms) are forecast to rise by around 59 per cent over this period—as average delays become longer, congestion more widespread and the proportion of freight and service vehicles increases. This is equivalent to a roughly 87 per cent increase in (metropolitan average) per capita congestion costs between 2005 and 2020. Figure S.1 displays how the unit congestion costs (cents per PCU-km) vary between the capital cities. The rightmost columns of Figure S.1 refer to weighted average values across all Australian metropolitan areas i.e., averaging the aggregate cost for the whole eight capitals across the aggregate VKT level.

16


Uni

t cos

ts (c

/km

)

0

2

4

6

8

10

12

14

Sydney

Melb

ourne

Brisb

ane

Adela

ide

Pert

h

Hobar

t

Dar

win

Can

berra

Metr

opolitan

avera

ge

20052020

Note: Costs here refer to avoidable social costs, and are based on the deadweight losses associated with the congestion levels. That is, these unit social costs refer to the estimated aggregate costs of delay, trip variability, vehicle operating expenses and motor vehicle emissions—associated with traffic congestion above the economic optimum level for the relevant networks—divided by the total VKT (in PCU-km) on the network.

Source: BTRE estimates.

Figure S.1 Average unit costs of congestion for Australian metropolitan areas—current and projected

As mentioned previously, both the complex nature of attempting to mathematically replicate the occurrence of urban congestion and the approximations inherent in the Bureau’s chosen aggregate methodology, lead to significant levels of uncertainty in the cost estimation process. As a guide to the level of approximation, and how wide the ranges are likely to be for the actual social costs of congestion, the report also contains the results of various sensitivity tests. Basically, these tests alter the input assumptions and the default parameter settings for the models to see how sensitive the final cost estimates are (to possible variation or uncertainty in such input values).

The following chart (figure S.2, adapted from figure 2.49 in the sensitivity section of the report) illustrates a typical range of variation in the cost estimates, depending on the exact input assumptions. The base case scenario values are plotted, along with both a high scenario (which is based on the highest likely population and economic growth over the projection period, coupled with minimal levels of future traffic peak spreading and significantly higher trip variability costs) and a low scenario (based on inputs of the lowest likely population and economic growth over the projection period, maximal levels of traffic peak spreading and with the proportion of trips assumed to not be time-sensitive set significantly higher).

A variety of other sensitivity tests for input parameter ranges and for model scenarios are also presented in the latter section of Part 2 of the report. One scenario estimated the rough effect on the modelled results of inputting acute (high and low) settings for urban public transit patronage and non-motorised transport participation—basically doubling non-car passenger-kilometre (pkm) task in the high scenario and setting it to zero in the low scenario, across the whole modelled time period (see figure 2.48). The model-response sensitivity scenario for a theoretical doubling of non-motorised and UPT travel (with approximately 12 per cent of aggregate urban pkm assumed to switch out of cars) shifted the model’s aggregate cost curve down on average by approximately 25 per cent. This impact, while substantial, is roughly equivalent to having each annual congestion cost value in the base case reached around 5 years later than in the timeline for base case projection.

There are a variety of modelling approaches typically used for estimating congestion costs—some studies (including this current

17

Summary

analysis) use aggregate methods, which are computationally straightforward but very approximate; others are based on detailed network models, which allow more precise traffic specifications but are much more data-intensive and difficult to adequately calibrate. The relative simplicity of aggregate methods means that, generally, they are readily calibrated. The current BTRE models have been calibrated against aggregate network performance data collected by the various State road authorities (including the annual statistics reported to the Austroads National Performance Indicators). Though this allows current congestion cost levels to be suitably benchmarked against actual on-road conditions, aggregate methods are very blunt instruments for projecting congestion occurrence. Traffic congestion projections are much more accurately performed using suitably calibrated network models and the BTRE is hopeful that the various jurisdictions will continue to work towards more detailed, and more nationally consistent, network modelling tools.

Though various Australian studies have tended to derive significantly varying congestion cost estimates, much of the variation can usually be explained by definitional differences (e.g. exactly which congestion effects are included in the costings) or by different input parameters or assumptions (e.g. underlying population or VKT growth rates, the dollar value of time for travel costs, or what free speed values are estimated for the network links). Probably the most significant factor in correctly determining the level of congestion costs is the assumed value of time used, where there is high uncertainty in how adequately standard values of time capture the worth that travellers actually attach to delays, and little data available on how the time values vary over different trip types and trip timings. If most urban trips in Australia are actually significantly less time sensitive than implied by the standard value of time, then the congestion cost estimates would have to be much lower than given in the current base case values.

In summary, and allowing for the uncertainty ranges discussed above, the costs imposed on Australian society by urban traffic congestion are likely to fall in the range of $5 to $15 billion for current levels—in terms of theoretically avoidable costs (i.e. if appropriate traffic management or pricing schemes were introduced so as to reduce traffic conditions to economically optimal levels)—with a median value of around $10 billion. This is likely to rise, under base case demand growth assumptions, to a level of between 10 and 30 billion

18


dollars by 2020, with a median projected value for avoidable social costs of congestion of around 20 billion dollars.

Though this range of likely values for national congestion impacts is quite large, even the lower bound values still indicate that urban traffic congestion imposes considerable costs on Australian society. Note also that the sensitivity scenarios have been framed primarily to demonstrate the dependence of the estimates on the input assumptions, rather than necessarily as plausible scenarios for the future in their own right. For example, given the wide range covered by the input variable settings for the two scenarios in figure S.2, it is highly likely that any realistic base case scenario to 2020 (run on the current BTRE model framework) would fall well between the upper and lower bounds displayed. Furthermore, most of the sensitivity analyses relate more to uncertainty over the exact valuation level for congestion costs (i.e. mainly concern definitional issues, such as what is the precise value of time felt by urban commuters or what is the most appropriate comparison speed to calculate average delays from) than to the projected trend in increasing cost levels. Practically all of the scenarios tested still exhibited between a twofold and threefold increase in projected congestion costs for the period of 2005 to 2020. Network modelling of individual Australian cities has typically derived projected cost trends with similar or stronger growth rates (e.g. see table 3.1, page 62, of VCEC 2006).

So, irrespective of the questions over exact dollar valuations raised by the sensitivity tests, the principal finding of this study remains: that, in the absence of improved congestion management, rising traffic volumes in the Australian capitals are likely to lead to escalating congestion impacts, such that the net social costs of congestion over the next 15 years (under a business-as-usual scenario) are likely to at least double.

19

Summary

Figure S.2 Trends in avoidable social costs of congestion—scenario results


$A b

illio

n

0

5

10

15

20

25

30

35

2020

2018

2016

2014

2012

2010

2008

2006

2004

2002

2000

1998

1996

1994

1992

1990

High scenario

Base case

Low scenario

20


21

Part 1—Traffic growth Trends in traffic and congestion in Australia’s capital cities

The types of growth that can cause problems for transport systems include:

• sudden growth such as export demand surges;

• unfunded growth where infrastructure struggles to cope; and

• growth in confined spaces.

The topics discussed in this section of the report are mainly concerned with the latter type of growth—where we examine the continuing traffic growth that underlies the increasing problems posed by urban congestion within the capital cities of Australia.

This part of the report looks at traffic growth trends, starting by examining the two major components of traffic growth—the passenger and freight tasks in our cities.

Growth in urban passenger demand

In the 60 years since the end of the Second World War, Australian cities have been transformed from fairly tightly knit core-and-spoke configurations, to sprawling suburban low-density configurations.

This transformation of urban land use has been accompanied and made possible by a rapid improvement and spread of the road system and an even more rapid expansion in car ownership (motor vehicles per person).

Figure 1.1 shows the 60 year growth in total passenger-kilometres (pkm) in our capital cities. The total task estimates for 1945 to 1976 are

22


derived from a combination of aggregate national data and, where available, State/Territory-specific travel trend information. The data from 1977 to 2004 represent the summation of estimates made by the BTRE for each of the eight capital cities using detailed jurisdiction-specific passenger data. The forecasts are derived on a primarily national aggregate basis—but using a methodology that allows for each State/Territory’s differing vehicle fleet and average personal travel characteristics.

As shown in figure 1.1, total travel in the urban areas of Australia has grown remarkably—ten-fold over 60 years. Almost all of that growth came from cars and ‘other’ road vehicles (mostly light commercial vehicles used for private purposes and motorcycles). As shown in Figure 1.2, private road vehicles (‘car plus other’ in the graph) represent about 90 per cent of the motorised passenger transport task in our capital cities. Urban public transport (UPT), though generally an important component of peak travel in many central business districts (CBDs), represents only a fairly minor share of today’s total urban passenger task. Moreover, UPT’s modal share has been remarkably constant since the early 1980s, when the long downward trend, following World War 2, in the transit share finally halted.

This picture is repeated when we consider the summed dataset for the eight capitals (see figures 1.3a and 1.3b). Today, UPT is growing, but for most jurisdictions only at rates comparable to the growth in travel by private road vehicles. This relative constancy of its share, combined with the dominance of car travel, allows us to analyse the growth in private road vehicle traffic in comparative isolation from the other forms of urban passenger transport.

Figure 1.1 Historical and projected urban passenger movement, Australian metropolitan total

Note: ‘Other’ primarily consists of non-business use of light commercial vehicles (LCVs)—with contributions from motorcycles, non-business use of trucks, and urban ferries.

Sources: BTRE (2003a, 2006a), BTRE estimates.

Figure 1.2 Historical and projected modal share, Australian metropolitan passenger travel

Note: ‘Other’ primarily consists of non-business use of light commercial vehicles (LCVs).


0

50

100

150

200

250 Other

Rail

Bus

Car

2020

1945

1948

1951

1954

1957

1960

1963

1966

1969

1972

1975

1978

1981

1984

1987

1990

1993

1996

1999

2002

2005

2008

2011

2014

2017

Billi

on p

asse

nger

–km

Base-caseprojections

Year ending June

Total metropolitan passenger task for Australia

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0Car + other

Rail

Bus

2020

1945

1948

1951

1954

1957

1960

1963

1966

1969

1972

1975

1978

1981

1984

1987

1990

1993

1996

1999

2002

2005

2008

2011

2014

2017


Prop

ortio

n of

tota

l met

ropo

litan

pas

seng

er-k

m

Year ending June

23

Part 1 | Traffic growth

0

50

100

150

200

250 Other

Rail

Bus

Car

2020

1945

1948

1951

1954

1957

1960

1963

1966

1969

1972

1975

1978

1981

1984

1987

1990

1993

1996

1999

2002

2005

2008

2011

2014

2017

Billi

on p

asse

nger

–km


Year ending June


0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0Car + other

Rail

Bus

2020

1945

1948

1951

1954

1957

1960

1963

1966

1969

1972

1975

1978

1981

1984

1987

1990

1993

1996

1999

2002

2005

2008

2011

2014

2017


Prop

ortio

n of

tota

l met

ropo

litan

pas

seng

er-k

m

Year ending June

24


Note that the values for bus passenger task and for bus VKT in this report (not only in the above two figures but also in all Table, Figure and text values to follow) refer to total commercial bus usage in metropolitan areas (i.e. all travel by commercial passenger vehicles with 10 or more seats). This includes not only the task carried by transit fleets, both privately-owned and government run, but also a lesser component of the total task due to charter/hire vehicles (which are often considerably smaller than a standard transit route bus). Therefore, the values given here for ‘UPT’ task (which sum heavy rail, light rail, bus and ferry passengers) are generally a little higher than if only the task carried by dedicated transit vehicles was included in the ‘bus’ estimates.

The main ‘drivers’ (or generators) behind the growth in total passenger travel in our cities (as well as behind the growth in travel by private road vehicles) are increases in population and increases in per capita travel (particularly as a result of rising per capita incomes). As is shown in figure 1.3c, there is a saturating relationship between increases in annual passenger-kilometres per person and per capita income. This relationship implies that saturation in per person travel could be mostly complete by 2020. Thereafter, population increase will tend to be the primary driver of increases in travel in our cities. Yet, at least until then, income increases will likely continue to add to per capita travel, and total passenger travel will grow at a faster rate than population.

The patterns in figures 1.3a to 1.3c are repeated in all the capital cities.

Sydney (see figures 1.4a to 1.4c) has experienced almost a doubling of the passenger transport task between 1977 and 2004. It has more urban public transport than the other capitals, due to large numbers of passengers taken relatively long distances by the rail system. However, even though the UPT mode share at 13 per cent is high, it has been virtually constant for many years (as in most other capitals). Per person travel is showing fairly clear signs of nearing saturation, at below 14 000 km per person. The main passenger modes are light motor vehicles (i.e. cars and personal use of LCVs), rail and bus.

Melbourne (figures 1.5a to 1.5c) has doubled the passenger transport task over the period. It has a lower UPT share (constant at about 8 per cent). Average travel per capita is above 14 000 km and still showing

25


some signs of further increase. The main modes are car, LCV, rail, light rail (trams), and bus.

Passenger travel in Brisbane (figures 1.6a to 1.6c) in 2004 is 2.5 times higher than in 1977. Again the UPT mode share is virtually constant, at about 8 per cent. Per person travel is showing signs of approaching stability at about 13 000 km per person per year. The main modes are car, LCV, rail and bus.

Adelaide (figures 1.7a to 1.7c) has shown slower growth in demand for passenger travel, with the total increasing 65 per cent from 1977 to 2004. UPT share has been drifting down over the period, but for recent years has been virtually constant at about 5 per cent. Per capita travel in Adelaide could possibly still be growing, and is currently around 13 000 km per person. The main modes are again car, LCV, rail and bus.

Passenger travel in Perth (figures 1.8a to 1.8c) slightly more than doubled from 1977 to 2004. The UPT share has come up from 5 ½ to 7 ½ per cent with the opening of the northern rail line and will probably approach 9 per cent with the opening of the southern line. But barring new rail lines, the UPT share is then likely to remain fairly constant (at a level below 10 per cent). Per person travel, currently around 13 000 km, is continuing to increase, and does not appear to have saturated yet with respect to per capita income levels. Once again, in order of task share, the main modes are car, LCV, rail and bus.

Hobart (figures 1.9a to 1.9c) is unique among the capitals for three reasons. First, it has shared with Adelaide slower growth in passenger travel (2004 at 1.75 times the 1977 level). Secondly, its total bus task has remained fairly constant over time (allowing for bus travel in charter/smaller buses, since it appears that Hobart’s UPT bus patronage has declined over time) so bus mode share has been declining in Hobart. Thirdly, travel per person started at the lowest level of all the capitals (8000 km per annum), and looks set to saturate at a relatively low level (in the order of 12 000 km). The main modes are car, LCV and bus.

Darwin (figures 1.10a to 1.10c) is also different from the other capitals, but for almost the opposite reasons to Hobart. It has registered the highest growth in passenger travel of any of the capitals (2004 at 2.7 times 1977 levels)—though from a comparatively low base. It has also seen mode share for bus travel increase over the period (although lately this has levelled off at a relatively high 10 per cent of total pkm).

26


Darwin’s per capita travel looks to be approaching saturation (between 11 000 and 12 000 km). The main modes are car, LCV and bus.

Canberra’s passenger travel (figures 1.11a to 1.11c) has more than doubled over the period. Mode share for buses is virtually constant at 6 per cent, and passenger travel per person appears to be practically saturated, at about 15 500 km per annum.

Part 2 of this paper provides detailed tables for annual travel trends (VKT) for the passenger vehicle fleets of each of the capital cities.

The procedure for translating this understanding of the patterns of passenger travel demand growth into forecasts of private vehicle traffic is documented after we examine the drivers behind the growth in the freight task in the cities. (For a schematic diagram brieftly summarising the main steps in the BTRE transport demand forecasting approach, see Appendix figure A.2.)

Figure 1.3a Historical trend in metropolitan passenger travel

Ferry

Bus

Light rail

RailMCs

LCVs

Car

Billi

on p

km

Eight capitals passenger–km task

1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

Motor Vehicles

UPT

Eight capitals mode share

1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

0

20

40

60

80

100

120

140

160

180

200

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

27


Figure 1.3b Historical trend in passenger mode share

Figure 1.3c Relationship of per capita travel to per capita income

Ferry

Bus

Light rail

RailMCs

LCVs

Car

Billi

on p

km

Eight capitals passenger–km task

1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

Motor Vehicles

UPT


1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

0

20

40

60

80

100

120

140

160

180

200

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

Pkm/Pop

Fitted trend

Eight capitals–pkm per person vs income per person, 1977 to 2004

15 000

14 000

13 000

12 000

11 000

10 000

9 000

8000

7000

6000

20 0

00

22 0

00

24 0

00

26 0

00

28 0

00

30 0

00

32 0

00

34 0

00

36 0

00

38 0

00

Aver

age

annu

al p

km p

er p

erso

n

Average income per person (GDP/Pop, 1998 dollars)

28


Figure 1.4a Historical trend in Sydney passenger travel

Figure 1.4b Historical trend in Sydney passenger mode share

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

Motor Vehicles

UPT


1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

0

10

20

30

40

50

60

70Ferry

Bus

Light rail

RailMCs

LCVs

Car

Billi

on p

km

Sydney passenger–km task

1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Motor Vehicles

UPT


Motor Vehicles

UPT

Sydney mode share

1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

Motor Vehicles

UPT


1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

0

10

20

30

40

50

60

70Ferry

Bus

Light rail

RailMCs

LCVs

Car

Billi

on p

km

Sydney passenger–km task

1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Motor Vehicles

UPT


Motor Vehicles

UPT

Sydney mode share

1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

29


Figure 1.4c Relationship of per capita travel to per capita income, Sydney

Figure 1.5a Historical trend in Melbourne passenger travel

Pkm/Pop

Fitted trend

Eight capitals–Pkm per person vs income per person, 1977 to 2004

15 000

14 000

13 000

12 000

11 000

10 000

9 000

8000

7000

6000

20 0

00

22 0

00

24 0

00

26 0

00

28 0

00

30 0

00

32 0

00

34 0

00

36 0

00

38 0

00

Aver

age

annu

al p

km p

er p

erso

n

Average income per person (GDP/Pop, 1998 dollars)

6000

8000

15 0

00

17 0

00

19 0

00

21 0

00

23 0

00

25 0

00

27 0

00

4000

2000

10 000

12 000

14 000

16 000

Aver

age

annu

al p

km p

er p

erso

n

Average income per person (PCFE/Pop)

Sydney–pkm person vs income per person, 1977 to 2004

Pkm/Pop

Fitted trend

0

10

20

30

40

50

60

Bus

Light rail

Heavy railMCs

LCVs

Car

Melbourne passenger–kms

Billi

on p

km

1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Motor Vehicles

UPT

1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

Melbourne mode share

30


Figure 1.5b Historical trend in Melbourne passenger mode share

Figure 1.5c Relationship of per capita travel to per capita income, Melbourne

0

10

20

30

40

50

60

Bus

Light rail

Heavy railMCs

LCVs

Car

Melbourne passenger–kms

Billi

on p

km

1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Motor Vehicles

UPT

1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

Melbourne mode share

6000

8000

4000

2000

10 000

12 000

14 000

16 000

14 0

00

16 0

00


18 0

00

20 0

00

22 0

00

24 0

00

26 0

00

28 0

00

Aver

age

annu

al p

km p

er p

erso

n

Melbourne–pkm person vs income per person, 1977 to 2004

Pkm/Pop

Fitted trend

31


Figure 1.6a Historical trend in Brisbane passenger travel

Figure 1.6b Historical trend in Brisbane passenger mode share

0

5

10

15

20

25Ferry

Bus

RailMCs

LCVs

Car

Brisbane passenger–kms

Billi

on p

km

1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Motor Vehicles

UPT

1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

Brisbane mode share

0

5

10

15

20

25Ferry

Bus

RailMCs

LCVs

Car

Brisbane passenger–kms

Billi

on p

km

1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Motor Vehicles

UPT

1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

Brisbane mode share

32


Figure 1.6c Relationship of per capita travel to per capita income, Brisbane

Figure 1.7a Historical trend in Adelaide passenger travel

14 000

12 000

10 000

8000

6000

4000

2000

12 0

00

14 0

00

16 0

00

18 0

00

20 0

00

22 0

00

24 0

00

26 0

00

Pkm/Pop

Fitted trend


Aver

age

annu

al p

km p

er p

erso

n

Brisbane–pkm person vs income per person, 1977 to 2004

0

2

4

6

8

10

12

14

16Bus

Light rail

Heavy railMCs

LCVs

Car

Billi

on p

km

1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

Adelaide passenger–kms

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

Motor Vehicles

UPT

Adelaide mode share

33


Figure 1.7b Historical trend in Adelaide passenger mode share

Figure 1.7c Relationship of per capita travel to per capita income, Adelaide

0

2

4

6

8

10

12

14

16Bus

Light rail

Heavy railMCs

LCVs

Car

Billi

on p

km

1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

Adelaide passenger–kms

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

Motor Vehicles

UPT

Adelaide mode share

14 000

12 000

10 000

8000

6000

4000

2000

Aver

age

annu

al p

km p

er p

erso

n

13 0

00

15 0

00

17 0

00

19 0

00

21 0

00

23 0

00

25 0

00


Pkm/Pop

Fitted trend

Adelaide–pkm person vs income per person, 1977 to 2004

34


Figure 1.8a Historical trend in Perth passenger travel

Figure 1.8b Historical trend in Perth passenger mode share

1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

Perth mode share

0

5

10

15

20

25Ferry

Bus

RailMCs

LCVs

Car

Billi

on p

km

Perth passenger–kms

1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Motor Vehicles

UPT

1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

Perth mode share

0

5

10

15

20

25Ferry

Bus

RailMCs

LCVs

Car

Billi

on p

km

Perth passenger–kms

1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Motor Vehicles

UPT

35


Figure 1.8c Relationship of per capita travel to per capita income, Perth

Figure 1.9a Historical trend in Hobart passenger travel

Aver

age

annu

al p

km p

er p

erso

n

Pkm/Pop

Fitted trend

7000

9000

11 000

13 000

15 000

10 0

00

12 0

00

14 0

00

16 0

00

18 0

00

20 0

00

22 0

00


24 0

00

26 0

00

Perth–pkm person vs income per person, 1977 to 2004

1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

Hobart mode share

Billi

on p

km

Hobart passenger–kms

1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

Motor Vehicles

UPT

0.00

0.50

1.00

1.50

2.00

2.50Bus

MCs

LCVs

Car

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

36


Figure 1.9b Historical trend in Hobart passenger mode share

Figure 1.9c Relationship of per capita travel to per capita income, Hobart

1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

Hobart mode share

Billi

on p

km

Hobart passenger–kms

1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

Motor Vehicles

UPT

0.00

0.50

1.00

1.50

2.00

2.50Bus

MCs

LCVs

Car

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Aver

age

annu

al p

km p

er p

erso

n


Hobart–pkm person vs income per person, 1977 to 2004

Pkm/Pop

Fitted trend

2000

4000

6000

8000

10 000

12 000

14 000

11 0

00

13 0

00

15 0

00

17 0

00

19 0

00

21 0

00

23 0

00

25 0

00

37


Figure 1.10a Historical trend in Darwin passenger travel

Figure 1.10b Historical trend in Darwin passenger mode share

1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

Darwin mode share

Billi

on p

km

Darwin passenger–kms

1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

Motor Vehicles

UPT

Bus

MCs

LCVs

Car

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

Darwin mode share

Billi

on p

km

Darwin passenger–kms

1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

Motor Vehicles

UPT

Bus

MCs

LCVs

Car

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

0.00

0.20

0.40

0.60

0.80

1.00

1.20

38


Figure 1.10c Relationship of per capita travel to per capita income, Darwin

Figures 1.11a Historical trend in Canberra passenger travel

Aver

age

annu

al p

km p

er p

erso

n


Darwin–pkm person vs income per person, 1986 to 2004

Pkm/Pop

Fitted trend

10 0

00

12 0

00

14 0

00

12 0

00

16 0

00

18 0

00

20 0

00

22 0

00

24 0

00

26 0

002000

4000

6000

8000

10 000

12 000

1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

Canberra mode share

Billi

on p

km

Canberra passenger–kms

1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

Motor Vehicles

UPT

Bus

MCs

LCVs

Car

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

0.00

1.00

2.00

3.00

4.00

5.00

6.00

39


Figure 1.11b Historical trend in Canberra passenger mode share

Figure 1.11c Relationship of per capita travel to per capita income, Canberra

1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

Canberra mode share

Billi

on p

km

Canberra passenger–kms

1977

1979

1985

1997

1981

1989

1987

1991

1993

1995

1999

2001

2003

1983

Motor Vehicles

UPT

Bus

MCs

LCVs

Car

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

0.00

1.00

2.00

3.00

4.00

5.00

6.00

Aver

age

annu

al p

km p

er p

erso

n


Canberra–pkm person vs income per person, 1977 to 2004

Pkm/Pop

Fitted trend

2000

4000

6000

8000

10 000

12 000

14 000

16 000

18 000

15 0

00

17 0

00

19 0

00

21 0

00

23 0

00

25 0

00

27 0

00

29 0

00

31 0

00

33 0

00

40


Urban freight growth

Growth in the freight task in the capital cities has been estimated in the BTRE’s recent publication Freight Measurement and Modelling (Report 112, BTRE 2006c—see Chapter 3 and Appendices II and III). The size of the task for the eight capitals combined is shown in figure 1.12. The task grew most rapidly in the 1970s and early 1980s, as real road freight rates declined sharply (see BTRE 2006b, Chapter 8).

Figure 1.12 also plots predicted values from a model of that growth, based on growth in GDP (income elasticity of 1.014) and reductions in real freight rates (real freight rate elasticity of -0.685).

A similar model, incorporating estimates for each capital, gives an income elasticity of 0.960 and a real freight rate elasticity of -0.781. The fit to the data for each city is shown in figure 1.13, together with forecasts from the model using GDP and real freight rates assumptions (detailed in Report 112). Table 1.1 gives the historical estimates of aggregate freight task (tonne-kilometres) for each capital, together with forecasts to 2020. It can be seen that the expected growth for the eight capitals in total is 3.1 per cent per year from 2003 to 2020, resulting in the projected eight-capital freight task growing by 70 per cent over that period.

Projected urban freight growth for most capitals is around that rate, except Hobart and Adelaide (which are estimated to be lower) and Brisbane and Perth (which are higher).

These forecasts of growth in total tonne-kilometre (tkm) task for each city are used to derive forecasts of growth in commercial vehicle traffic using methods described in the section after next.

41


Table 1.1 Capital city road freight task—trends and projections

(billion tonne-kilometres)

Year Syd Mel Bne Adl Per Hob Drw Cbr 8 Caps

1971 2.82 1.96 0.62 0.67 0.78 0.11 0.04 0.10 7.111972 2.98 2.09 0.68 0.71 0.84 0.13 0.05 0.11 7.581973 3.14 2.21 0.75 0.75 0.89 0.14 0.05 0.11 8.041974 3.29 2.34 0.81 0.79 0.94 0.15 0.05 0.12 8.501975 3.44 2.47 0.88 0.83 1.00 0.16 0.06 0.13 8.961976 3.59 2.60 0.95 0.87 1.05 0.18 0.06 0.13 9.431977 3.85 2.83 1.12 0.94 1.15 0.19 0.06 0.14 10.281978 4.10 3.06 1.30 1.00 1.24 0.21 0.06 0.16 11.131979 4.34 3.29 1.49 1.07 1.34 0.23 0.06 0.17 11.981980 4.53 3.47 1.66 1.05 1.41 0.24 0.09 0.18 12.631981 4.72 3.66 1.84 1.02 1.49 0.24 0.13 0.18 13.281982 4.90 3.84 2.03 0.98 1.56 0.25 0.17 0.19 13.931983 5.05 4.00 2.11 1.02 1.60 0.27 0.18 0.20 14.441984 5.20 4.16 2.20 1.06 1.64 0.29 0.19 0.21 14.951985 5.34 4.32 2.29 1.10 1.68 0.32 0.20 0.21 15.461986 5.64 4.64 2.43 1.17 1.79 0.33 0.20 0.23 16.421987 5.92 4.96 2.57 1.24 1.90 0.33 0.21 0.24 17.371988 6.21 5.28 2.72 1.31 2.01 0.34 0.21 0.25 18.331989 6.37 5.51 2.80 1.35 2.09 0.34 0.22 0.26 18.941990 6.53 5.73 2.89 1.38 2.18 0.35 0.23 0.27 19.551991 6.69 5.96 2.97 1.42 2.27 0.35 0.24 0.28 20.171992 6.99 6.30 3.13 1.48 2.38 0.35 0.24 0.27 21.141993 7.29 6.64 3.30 1.55 2.50 0.34 0.24 0.26 22.121994 7.58 6.99 3.47 1.62 2.62 0.33 0.23 0.26 23.101995 7.88 7.34 3.64 1.68 2.74 0.33 0.23 0.25 24.081996 8.19 7.67 3.86 1.76 2.86 0.31 0.24 0.25 25.121997 8.49 8.00 4.08 1.83 2.98 0.29 0.25 0.24 26.171998 8.80 8.33 4.31 1.91 3.10 0.27 0.26 0.24 27.221999 8.95 8.61 4.64 1.93 3.17 0.27 0.29 0.24 28.102000 9.35 9.20 4.99 2.01 3.40 0.28 0.26 0.25 29.742001 9.62 9.45 5.18 2.04 3.52 0.29 0.22 0.24 30.562002 10.02 10.06 5.50 2.15 3.75 0.31 0.19 0.25 32.222003 10.39 10.28 5.71 2.25 3.87 0.32 0.19 0.26 33.27

Average annual growth rate (per cent)

1985–2003 3.8 4.9 5.2 4.1 4.7 0.0 -0.3 1.2 4.3

42


Table 1.1 Capital city road freight task—trends and projections (continued)

(billion tonne-kilometres)

Year Syd Mel Bne Adl Per Hob Drw Cbr 8 Caps

2003act 10.39 10.28 5.71 2.25 3.87 0.32 0.19 0.26 33.272003pred 10.84 10.21 5.58 2.23 3.91 0.33 0.20 0.28 33.592004 11.22 10.58 5.81 2.30 4.07 0.34 0.21 0.29 34.822005 11.64 11.00 6.08 2.37 4.24 0.35 0.23 0.30 36.202006 12.08 11.43 6.35 2.45 4.41 0.36 0.24 0.31 37.622007 12.54 11.86 6.64 2.52 4.60 0.37 0.25 0.32 39.092008 12.93 12.23 6.89 2.58 4.77 0.38 0.26 0.33 40.372009 13.32 12.61 7.15 2.65 4.92 0.39 0.27 0.34 41.652010 13.71 12.99 7.41 2.71 5.09 0.40 0.28 0.35 42.942011 14.11 13.36 7.68 2.77 5.27 0.41 0.29 0.36 44.262012 14.52 13.75 7.96 2.85 5.44 0.41 0.29 0.37 45.592013 14.94 14.15 8.23 2.91 5.62 0.42 0.30 0.38 46.952014 15.34 14.54 8.51 2.97 5.79 0.43 0.31 0.39 48.292015 15.76 14.93 8.79 3.04 5.97 0.44 0.32 0.40 49.652016 16.16 15.31 9.07 3.10 6.14 0.45 0.33 0.41 50.972017 16.58 15.71 9.36 3.16 6.33 0.45 0.34 0.42 52.362018 17.04 16.14 9.68 3.23 6.52 0.46 0.35 0.43 53.862019 17.43 16.51 9.96 3.29 6.70 0.47 0.37 0.44 55.172020 17.85 16.92 10.27 3.35 6.89 0.48 0.38 0.45 56.60

Average annual growth rate (per cent)

2003 pred—2020 3.0 3.0 3.7 2.4 3.4 2.2 3.8 2.8 3.1

Source: BTRE (2006b).

Figure 1.12 Historical and model fitted data for aggregate freight task

Source: BTRE (2006b).

0

2

4

6

8

10

12

14

16

18

20

1971 1971 19712020 2020

19712020

19712020

19712020

19712020

19712020 2020

billi

on to

nne-

kilo

met

res

MelbourneSydney

City Freight

City Freight—Predicted

Brisbane

Perth

Hobart Darwin Canberra

Adelaide

2001

2003

1999

1997

1995

1993

1991

1989

1987

1985

1983

1981

1979

1977

1975

1973

1971

0

5

10

15

20

25

30

35

40City Freight

City Freight —Predicted

billi

on to

nne-

kilo

met

res

43


Figure 1.13 Cross section, time-series data and model fit for city freight tasks

Source: BTRE (2006c).

Car traffic growth

Given the comparative constancy of the expected UPT mode share in each city (including Perth after the addition of the southern railway), essentially one can forecast car traffic growth relatively independently.

A simplifying framework for explaining car traffic (vehicle kilometres travelled or VKT) is the following:

Car traffic = Car travel per person × Population

The advantage of this formulation is that, for Australia, it turns out that car travel per person has a simple relationship to economic activity levels. The trend in per capita car travel (kilometres per person) in Australia has in general been following a logistic (saturating) curve against real per capita income—measured here by real Gross Domestic Product (GDP) per person (see figure 1.14).

0

2

4

6

8

10

12

14

16

18

20

1971 1971 19712020 2020

19712020

19712020

19712020

19712020

19712020 2020

billi

on to

nne-

kilo

met

res

MelbourneSydney

City Freight

City Freight—Predicted

Brisbane

Perth

Hobart Darwin Canberra

Adelaide

2001

2003

1999

1997

1995

1993

1991

1989

1987

1985

1983

1981

1979

1977

1975

1973

1971

0

5

10

15

20

25

30

35

40City Freight

City Freight —Predicted

billi

on to

nne-

kilo

met

res

44


Figure 1.14 Historical trend in annual per capita passenger vehicle travel versus real Australian income per capita

Source: BTRE (2002, 2003a, 2005), BTRE Estimates.

Here, then, we have the basis for understanding the relationship between car traffic and economic development. As incomes per person increase, personal car travel per person has also tended to increase, but at a slowing rate over time. In other words, more car travel is attractive as incomes rise, but there reaches a point where further increases in per capita income elicit no further demand for car travel per capita. However, even after the virtual saturation of per capita km travelled to per capita income growth, total car traffic continues to respond (in an essentially one-to-one relationship now) to population growth (that other component of aggregate economic activity levels).

Note that even though personal passenger travel exhibits a saturating trend over time, there is, as yet, no sign of approaching saturation in per capita freight movement in Australia. For illustrative purposes, a graph of national per capita passenger and freight movement curves—relative to per capita income—is provided in the Appendix, as figure A.1.

Once again, our formula for understanding the relationship between car traffic growth and economic growth is:

Car traffic = Car travel per person × Population

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

8.0

9.0

10.0Logistic fitted curve

Actual personal travel (1950–2005)

Thou

sand

km

of t

rave

l per

per

son

Index of real income per capita

0 5 15 2010

Trend in per capita Australian vehicle travel

5025 30 35 40 45

45


The assumed base case rate of GDP growth of around 2.7 per cent per annum over the 15 years from 2005 to 2020 (Treasury 2002) implies that Australia-wide per capita car travel should level out at around 8900 kilometres per person by 2020—about a 6 per cent increase on 2005 (and about 10–12 per cent over the somewhat below trend 2001–2003 levels). After 2020, growth coming from this first term in the equation will probably effectively cease.

There is still the growth in car travel resulting from population growth to consider. The two main sources of population growth are natural increase and immigration. The contribution each has made to population growth over approximately the last 40 years is shown in Figure 1.15 (where the two components have been stacked). The average growth rate (of both components) has tended to decline over time.

The Australian Bureau of Statistics (ABS) has previously produced several scenarios for population growth—see www.abs.gov.au for details—projecting national population to be between about 22 million and 24 million people by 2020. The following aggregate analysis example uses population projections based on the trends to 2020 of the ABS Series III projections (e.g. ABS 2001) and the recent mid-range Series B long-term projections (ABS 2005). ‘Series III’ projections assume a net immigration level of about 70 000 persons per year and a further decline in the rate of natural increase (due to a fairly rapid ageing of the population, coupled with a fairly low fertility rate).

The ABS population projections are also available for each of the Australian capital cities (see the Appendix for tabulated population projection values based on the ABS mid-range values). The following example values (e.g. table 1.2) are derived using a rough aggregate approach—to demonstrate the overall trend picture (where final values from the detailed modelling and projection process—that allows for demographic characteristics and state-specific vehicle fleet attributes—are given within tables in Part 2 of this paper). So, for simplicity, if we use national car travel per person percentage increases (from figure 1.14) and the capital city population projections, we obtain the VKT values in table 1.2, for this illustration of the car traffic projection method. For example, using a year 2005 baseline, the national (percentage) projected increase in car travel per person is roughly:

46


(8.87-8.38)/8.38 = 5.8%.

Increasing Sydney’s per capita travel (at approximately 7.47 thousand km VKT per person in 2005) by this amount gives an illustrative projected 2020 level of about 7.91 thousand km per person. Multiplying this by Sydney’s projected 2020 population of around 5.1 million gives (illustrative) projected 2020 Sydney car VKT of about 40.3 billion km.

It should be noted that the national level of VKT per person is slightly higher than the metropolitan average, but for this example it is assumed that the latter will saturate in a like manner to the national total.

Figure 1.15 Historical trend in components of Australian population increase

Sources: ABS (2001b).

Popu

latio

n gr

owth

Net immigration

Natural increase

0.0%

0.5%

1.0%

1.5%

2.0%

2.5%

1961

1966

1971

1976

1981

1986

1991

1996

2001

47


Table 1.2 Example car traffic projections for Australian cities

City 2005 2020 Percent change

Car Population Total Car Car Population Total Car 2005– VKT/ (‘000) VKT VKT/ (‘000) VKT 2020 Person (million) Person (million) (‘000) (‘000)

Sydney 7.47 4 382 32 715 7.91 5 103 40 340 23%Melbourne 8.60 3 682 31 651 9.10 4 143 37 702 19%Brisbane 7.30 1 780 12 996 7.73 2 233 17 264 33%Adelaide 7.96 1 135 9 032 8.43 1 195 10 069 11%Perth 7.58 1 505 11 411 8.03 1 835 14 731 29%Hobart 7.59 194 1 474 8.04 192 1 540 5%Darwin 6.33 99 630 6.71 130 871 38%Canberra 9.55 328 3 133 10.11 362 3 663 17%

Metro 7.86 13 106 103 041 8.32 15 193 12 6475 23%

Rest of Australia 9.42 7 244 68 265 9.98 8 049 80 308 18%

Total Aust. 8.42 20 350 171 306 8.91 23 241 207 154 21%

Note: The projected Australian aggregate level per cent increase (of 8.38 thousand km per person in 2005 to near saturation at 8.87 by 2020) is assumed here to apply to each city. At the level of the 8 capitals, the increase from car travel per person is about 6%, and from population 16%. The overall increase in Australia Metro car traffic is therefore likely to be of the order of (1.06 × 1.16—1) = about 23% over the 15 years.


The average increase in car traffic in Australian capital cities is projected to be in the order of 23 per cent (close to the Sydney and Melbourne levels of growth, with the most significant in Brisbane, because of its high population growth). Even with a proportion of the total growth occurring at the city fringes, this still implies substantial increases in the level of car traffic on our current city networks.

As mentioned above, this procedure (table 1.2) is a simplification of the final process of forecasting car traffic growth in the cities. The final forecasts for each capital’s car VKT are given in tables in Part 2. For now, we turn to forecasting the rest of the traffic stream.

Bus and motorcycle traffic growth

Buses and motorcycles form a small part of passenger vehicle traffic in Australian cities (excluding for the moment the traffic contribution

48


of LCVs and trucks). In most of our cities, they account for only 1–3 per cent of the total VKT by non-business vehicles (i.e. cars routinely account for 97–98 per cent of passenger vehicle traffic). This car share has practically saturated. For the bus and motorcycle VKT shares, bus travel has tended, on average, to grow slightly over the last decade and a half, while the motorcycle share reduced during the 1990s—though often with a relatively high year to year variability. Motorcycle use appears to be currently growing again, after many years of decline, and could possibly see its share increase in the future, especially if their manoeuvrability in traffic becomes more attractive as congestion levels grow. Equivalently, if significant numbers of drivers find the hassle of coping with congested driving conditions not worthwhile, as future congestion becomes more widespread, bus patronage could also gain from some possible modal shift, though from a purely delay point of view, standard buses will not generally show any benefits over car travel.

BTRE projections of non-car passenger vehicle travel are based on competitiveness models (using generalised costs, which take account of travel time spent as well as direct expenses such as fuel prices and fares). The BTRE base-case projections have the bus and motorcycle share of passenger VKT as remaining essentially constant (only increasing marginally by 2020 from the current combined VKT share of about 1.7 per cent of national metropolitan passenger vehicle kilometres travelled).

As for the metropolitan car VKT projections, (base case or business-as-usual) forecasts for bus and motorcycle traffic are presented in Part 2 tables.

Truck traffic growth

The basic mechanism generating truck traffic can be expressed as follows:

Truck Traffic (VKT) = Road Freight Task / Average Load per Truck

In other words, a certain level of truck kilometres is performed in order to carry out the freight task in each city, where the number of vehicles travelling is determined by the average load of those

49


vehicles. The level of truck traffic (in total VKT terms) can equivalently be derived from the product of numbers of vehicles times the yearly average VKT they each perform.

The influences of the economy and technological shifts (in improving logistical operation or truck size) can then be illustrated as below:

Total freight vehicle traffic (VKT per annum) =

The main influence of economic development is through increases in the freight task. In the section above on urban freight, tkm growth was found to react proportionally to the growth rate of the economy—at about 1.0 times economic growth. While this relationship cannot continue indefinitely, there are no signs yet of saturation in Australian truck freight use per person (as there are in car travel per person). Similarly, there are no signs of saturation in current levels of United States truck freight per person, and American levels of road freight per person are already much higher than those in Australia. The other influence on the demand for urban freight transport is the real freight rate. Real road freight rates in Australia have fallen dramatically since 1965, mainly driven by the progressive introduction of larger articulated vehicles, but also by technological change which has made possible lighter vehicles, improved terminal efficiencies etc. Real freight rates fell 45 per cent from 1965 to 1990, and then another 3 per cent in the 1990s (BTRE 2002b). In the section above on urban freight, demand was seen to increase 0.7 to 0.8 per cent with a 1.0 per cent reduction in real freight rates.

The other influence of technological change is direct. For example, the same weight-reducing technological change that lowers freight rates also makes possible direct increases in average loads.

Number of × Average km = Road Freight Task (tkm) / Average load per truck(t) Vehicles per vehicle

Economicdevelopment

Falls in freight rates

Technical change

Shift to larger truck types

(–)

(–)

(+)

(+)

(+) (+) (+) (+)

(+)

(+)

50


However, the main influence on average loads has been the continuing shift to the larger articulated vehicles. This directly increases average load and serves to reduce the overall number of trucks on the road (i.e. from the level of truck traffic that would have been required at the lower average loadings).

Overall, then, the effects of economic development and associated technical change can be summarised as follows:

1. As the economy grows, the urban road freight task grows just as quickly.

2. The shift to larger vehicles makes possible larger loads and therefore less traffic (albeit composed of larger vehicles), but at the same time makes possible lower real freight rates which causes additional demand for freight transport.

3. General technological change has a similar ‘double-edged’ effect on truck traffic.

The projections in Table 1.1 have already provided the forecasts of the aggregate freight task in tonne-kilometres for each capital. These are turned into traffic forecasts in two steps:

1. The task is split into vehicle types by assumptions about the trend in vehicle type share. The result is projections of tonne-kilometres performed by LCVs, rigid trucks and articulated trucks.

2. The change in average loads per vehicle type is projected, based on past trends and assessments of likely technological or industrial changes.

The result is forecasts for each capital of the VKT performed by LCVs, rigid trucks and articulated trucks.

Projections of total traffic for Australian cities

The forecast growth in total traffic by vehicle type in the various cities is given in detail in the next section of the paper (see tables 2.1 to 2.9 for the VKT projections by vehicle type for each capital city). The following chart (figure 1.16) provides a summary of these projected trends. Table 2.10 weights these VKT projections (in km) by vehicle

51


type to more accurately reflect the traffic-impedance value of each vehicle class. Typical weights versus a passenger car (equal to one), for example, are two for rigid trucks and buses, and three for a six-axle articulated truck. The weighted values are commonly called passenger car equivalent units (PCUs). Figure 1.17 plots the projected VKT trend (aggregate PCU-km, using the BTRE base case scenario) for the Australian capital cities. (For a schematic diagram briefly summarising the demand projection process see Appendix figure A.2.)

Variations in forecast city growth rates for VKT (e.g. the high growth in Brisbane, Perth and Darwin, and the low growth in Hobart) are due mainly to variations in projected population growth. The average metropolitan growth in traffic (total PCU terms, across the eight capitals) is projected to be about 37 per cent over the 15 years from 2005 to 2020. (Though it should be noted that there might be several reasons for traffic growth to be constrained below these forecasts, possibly through a stronger than expected demand response to increasing levels of congestion, through higher than expected changes in demand patterns due to traffic control measures or modal shifts, or from higher than projected fuel prices).

Cars continue to be the largest component of the traffic stream. Their forecast growth of about 24 per cent is, as we have seen, composed of around 6 per cent growth coming from the effect of rising income levels on per person travel, and the rest from the projected increase in population of the eight capital cities. Buses and motorcycles continue to be a small part of the traffic stream.

Articulated truck use is projected to continue growing strongly, but their overall vehicle numbers are relatively small. However, LCVs are a substantial, and quickly growing, part of the traffic stream (with forecast 2005 to 2020 VKT growth of 90 per cent in the base case). Growth in LCV use has averaged between 3 and 4 per cent per annum for well over 20 years, and the base case essentially continues this trend to 2020, with continued (projected) economic growth leading to continued VKT growth.

As mentioned previously, this relatively high level of commercial traffic growth is predicated on the assumption that there will be no decoupling of activity in the freight and service sectors from overall income trends (i.e. GDP per person) during the projection period. If such decoupling does occur in the future (as has already become apparent for the

52


passenger sector), VKT growth for LCVs and trucks will be expected to decelerate, but the evidence so far suggests that such a saturating trend in per capita freight movement is unlikely in the short to medium term (e.g. see Appendix figure A.1). It is primarily the strong projected annual growth in LCV travel that substantially lifts the forecast growth in total metropolitan traffic (PCU-km) between 2005 and 2020 to 37 per cent, from the 24 per cent level for cars. (Though note that even if the current almost exponential growth in commercial VKT slows to, say, a linear growth trend midway during the projection period, then the overall PCU-km growth over the period would not be radically different, changing from around 37 per cent to the order of 33–35 per cent).

PCU values are generally the best ones for gauging the congestion potential of the total traffic stream into the future. Looking at Figure 1.17, two things are clear:

1. The growth in total traffic (in PCU terms) is expected to be approximately linear over the full period from 1990 to 2020.

2. The projected growth rate means that the same absolute volume of traffic (in PCU terms) will likely be added to our capital city roads in the next 15 years as was added in the past 15.

The growth in traffic in the past 15 years has resulted in additional congestion, but the increase has been moderated by three main factors:

1. Significant additions to capacity (in the form of freeways, tunnels etc.) for many cities;

2. Increasing intelligence built into the road network (e.g. loops in the road controlling intersection lights); and

3. Peak spreading.

None of these three factors will cease operating in the next 15 years, but given the extent of their implementation in many areas, obtaining substantial additions to their current moderating influence will likely pose a challenge for some jurisdictions.

The next part of this report discusses the implications of the large projected increases in traffic in our cities for increases in congestion, and in the associated social costs. Suffice to say, increases in traffic of the size foreseen here will have major implications for mobility and amenity in our cities.

53


Figure 1.16 Projected travel by motor vehicles, Australian metropolitan total


Figure 1.17 Total projected traffic for Australian capital cities


Billi

on k

ilom

etre

s tr

avel

led

National metropolitan VKT

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

0

20

40

60

80

100

120

140

160

180Motorcycles

Buses

Articulated trucksRigid trucks

LCVs

Car

2018

2020

0

20

40

60

80

100

120

140

160

180

200

Billi

on P

CU

–km

trav

elle

d

National metropolitan VKT – all vehicles (car equivalents)

Hobart

Perth

AdelaideBrisbane

Melbourne

Sydney

Canberra

Darwin

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

Billi

on k

ilom

etre

s tr

avel

led


1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

0

20

40

60

80

100

120

140

160

180Motorcycles

Buses


LCVs

Car

2018

2020

0

20

40

60

80

100

120

140

160

180

200

Billi

on P

CU

–km

trav

elle

d

National metropolitan VKT – all vehicles (car equivalents)

Hobart

Perth

AdelaideBrisbane

Melbourne

Sydney

Canberra

Darwin

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

55

Part 2–Social costs of congestionEstimation of Avoidable Social Costs for Australian Capital Cities

Increasing traffic trends underlying congestion

Motor vehicle travel within Australian cities has grown enormously over the last 60 years–with current levels of urban kilometres travelled by passenger cars being over 15 times greater than at the start of the 1950s (see figure 2.1 for the long-term trends in urban passenger movement since World War II). Furthermore, as also shown in figure 2.1 (under ‘base case’ projection assumptions), metropolitan vehicle travel in Australia is expected to continue to grow appreciably over the next decade and a half.

Figure 2.1 Historical and projected urban passenger movement, Australian metropolitan total

Note: ‘Other’ primarily consists of non-business use of light commercial vehicles (LCVs)–with contributions from motorcycles, non-business use of trucks, and urban ferries.

Sources: BTRE (2003a, 2005), BTRE estimates.

Billi

on p

asse

nger

–km


0

50

100

150

200

250Other

Rail

Bus

Car

2020

1945

1948

1951

1954

1957

1960

1963

1966

1969

1972

1975

1978

1981

1984

1987

1990

1993

1996

1999

2002

2005

2008

2011

2014

2017


0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0Car + other

Rail

Bus


2020

1945

1948

1951

1954

1957

1960

1963

1966

1969

1972

1975

1978

1981

1984

1987

1990

1993

1996

1999

2002

2005

2008

2011

2014

2017

56


Figures 2.1 and 2.2 nicely summarise the overall modal trends experienced in Australian metropolitan travel over the long term, so they have been reproduced here from Part I of the paper, to introduce our discussion of the impacts of rising traffic levels and consequent congestion effects.

The main ‘drivers’ of growth in overall transport demand have traditionally been increases in population and average income levels. As demonstrated in Part I of this study, future increases in Australian urban passenger travel are likely to be more dependant on the rate of population increase and less dependent on increases in general prosperity levels. The growth rate in passenger-kilometres (pkm) per person has reduced in recent years, especially compared with the some high growth periods in the past (such as between the 1950s and the 1970s).

Basically, as income levels and motor vehicle affordability have increased over time, average travel per person has increased. However, there are limits to how far such growth can continue. Eventually, people are spending as much time on daily travel as they are willing to commit and are loath to spend any more of limited time budgets on even more travel, even if average income levels continue to increase. So growth in per capita personal travel is likely to be lower in the future than for the long-term historical trend. However, this decoupling of income level trends from personal travel trends is not apparent in the current freight movement trends. Tonne-kilometres performed per capita are still growing quite strongly, and even though the freight trend curve could possibly also exhibit slowing growth over the longer-term, there is no saturating tendency evident yet (see Appendix figure A.1). Growth in freight and service vehicle traffic is therefore projected to be substantially stronger than for passenger vehicles, over the next decade and a half.

The BTRE base case (or ‘business-as-usual’) projections of Australian transport use are derived from forecasts of population and income levels for each of the relevant regions—allowing for projected trends in fuel prices and other travel expenses (such as fares or vehicle purchase prices) using a variety of aggregate demand models and/or modal competition models. For some background material on the base-case projection process see: BTRE Report 107 (BTRE 2002a), Greenhouse Gas Emissions from Australian Transport: Base Case

57

Part 2–Social costs of congestion

Projections to 2020 (BTRE 2006a), and Urban Pollutant Emissions from Motor Vehicles: Australian Trends To 2020 (BTRE 2003a).

Under the base-case scenario settings, the modal share of urban public transport is not projected to vary significantly, from current levels, over the next decade and a half (see figure 2.2). Public transit patronage has reasonably strong growth in the base case (stronger that for total car use) but since over 90 per cent of the total pkm task is done by light vehicles, the portion of mode share that cars lose to buses and rail over the projection period does not make much of a change to their level of dominance. The BTRE projects private travel volumes (averaged across the eight capital cities) to increase by about 1.7 per cent per annum over the period of 2000 to 2020, with a stronger growth trend for the commercial road sector (business kilometres expected to increase by around 3.5 per cent per annum, 2000 to 2020). High levels of traffic and traffic growth lead to significant levels of urban congestion—especially in peak travel periods—which imposes considerable costs on those affected by delays, increased fuel consumption and increased air pollution.

Figure 2.2 Historical and projected modal share, Australian metropolitan passenger travel

Note: ‘Other’ primarily consists of non-business use of light commercial vehicles (LCVs).


Billi

on p

asse

nger

–km


0

50

100

150

200

250Other

Rail

Bus

Car

2020

1945

1948

1951

1954

1957

1960

1963

1966

1969

1972

1975

1978

1981

1984

1987

1990

1993

1996

1999

2002

2005

2008

2011

2014

2017


0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0Car + other

Rail

Bus


2020

1945

1948

1951

1954

1957

1960

1963

1966

1969

1972

1975

1978

1981

1984

1987

1990

1993

1996

1999

2002

2005

2008

2011

2014

2017

58


The total vehicle kilometres travelled (VKT) estimates from the base case projections (displayed previously in figure 1.16) are shown in figure 2.3. (For a schematic diagram briefly summarising the demand projection process see Appendix figure A.2.)

As shown in table 2.1, total metropolitan VKT (all vehicle types) is forecast, in the BTRE base case, to increase by about 34 per cent between 2005 and 2020.

Figure 2.3 Historical and projected travel by motor vehicles, Australian metropolitan total

Sources: BTRE estimates.

Australian cities not only vary greatly in size and design (and consequently, in current traffic congestion levels), but also have varying levels of population growth (and consequently, expected congestion growth rates). Figures 2.4 to 2.19 present BTRE projections of motor vehicle traffic for the State and Territory capitals to 2020. Figure 2.11 gives total annual kilometres travelled in passenger car equivalent units (PCU-km).

As shown in table 2.10, total metropolitan traffic (in PCU-weighted VKT terms) is projected to increase by close to 37 per cent between 2005 and 2020 (using base case input assumptions to the BTRE transport demand models).

Billi

on k

ilom

etre

s tr

avel

led


1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

0

20

40

60

80

100

120

140

160

180Motorcycles

Buses


LCVs

Car

2018

2020

Billi

on k

ilom

etre

s tr

avel

led

National metropolitan VKT – passenger cars

Hobart

Perth

AdelaideBrisbane

Melbourne

Sydney

Canberra

Darwin

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

0

20

40

60

80

100

120

140

59


Figure 2.4 Total projected car traffic for Australian capital cities


Figure 2.5 Total projected LCV traffic for Australian capital cities


Billi

on k

ilom

etre

s tr

avel

led


1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

0

20

40

60

80

100

120

140

160

180Motorcycles

Buses


LCVs

Car

2018

2020

Billi

on k

ilom

etre

s tr

avel

led

National metropolitan VKT – passenger cars

Hobart

Perth

AdelaideBrisbane

Melbourne

Sydney

Canberra

Darwin

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

0

20

40

60

80

100

120

140

Billi

on k

ilom

etre

s tr

avel

led

National metropolitan VKT – LCV’s

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

0

5

10

15

20

25

30

35

Hobart

Perth

AdelaideBrisbane

Melbourne

Sydney

Canberra

Darwin

Billi

on k

ilom

etre

s tr

avel

led

National metropolitan VKT – rigid trucks

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

Hobart

Perth

AdelaideBrisbane

Melbourne

Sydney

Canberra

Darwin

0

1

2

3

4

5

6

60


Figure 2.6 Total projected rigid truck traffic for Australian capital cities


Figure 2.7 Total projected articulated truck traffic for Australian capital cities


Billi

on k

ilom

etre

s tr

avel

led

National metropolitan VKT – LCV’s

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

0

5

10

15

20

25

30

35

Hobart

Perth

AdelaideBrisbane

Melbourne

Sydney

Canberra

Darwin

Billi

on k

ilom

etre

s tr

avel

led

National metropolitan VKT – rigid trucks

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

Hobart

Perth

AdelaideBrisbane

Melbourne

Sydney

Canberra

Darwin

0

1

2

3

4

5

6

Billi

on k

ilom

etre

s tr

avel

led

National metropolitan VKT – Articulated trucks

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

Hobart

Perth

AdelaideBrisbane

Melbourne

Sydney

Canberra

Darwin

0.0

0.5

1.0

1.5

2.0

2.5

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

Billi

on k

ilom

etre

s tr

avel

led

National metropolitan VKT – Buses

Hobart

Perth

AdelaideBrisbane

Melbourne

Sydney

Canberra

Darwin

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

61


Figure 2.8 Total projected bus traffic for Australian capital cities


Figure 2.9 Total projected motorcycle traffic for Australian capital cities


Billi

on k

ilom

etre

s tr

avel

led

National metropolitan VKT – Articulated trucks

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

Hobart

Perth

AdelaideBrisbane

Melbourne

Sydney

Canberra

Darwin

0.0

0.5

1.0

1.5

2.0

2.5

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

Billi

on k

ilom

etre

s tr

avel

led

National metropolitan VKT – Buses

Hobart

Perth

AdelaideBrisbane

Melbourne

Sydney

Canberra

Darwin

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

Billi

on k

ilom

etre

s tr

avel

led

National metropolitan VKT – Motorcycles

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

Hobart

Perth

AdelaideBrisbane

Melbourne

Sydney

Canberra

Darwin

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

0

20

40

60

80

100

120

140

160

180

Billi

on k

ilom

etre

s tr

avel

led

National metropolitan VKT – All vehicles

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

Hobart

Perth

AdelaideBrisbane

Melbourne

Sydney

Canberra

Darwin

62


Figure 2.10 Total projected traffic for Australian capital cities–VKT


Figure 2.11 Total projected traffic for Australian capital cities–PCU VKT


Billi

on k

ilom

etre

s tr

avel

led

National metropolitan VKT – Motorcycles

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

Hobart

Perth

AdelaideBrisbane

Melbourne

Sydney

Canberra

Darwin

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

0

20

40

60

80

100

120

140

160

180

Billi

on k

ilom

etre

s tr

avel

led

National metropolitan VKT – All vehicles

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

Hobart

Perth

AdelaideBrisbane

Melbourne

Sydney

Canberra

Darwin

Billi

on k

ilom

etre

s tr

avel

led

National metropolitan VKT – All vehicles (car equivalents)

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

0

20

40

60

80

100

120

140

160

180

200

Hobart

Perth

AdelaideBrisbane

Melbourne

Sydney

Canberra

Darwin

0

10

20

30

40

50

60

Billi

on k

ilom

etre

s tr

avel

led

Sydney VKT

Motorcycles

Buses


LCVs

Car

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

63


Figure 2.12 Total projected VKT for Sydney


Figure 2.13 Total projected VKT for Melbourne


Billi

on k

ilom

etre

s tr

avel

led

National metropolitan VKT – All vehicles (car equivalents)

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

0

20

40

60

80

100

120

140

160

180

200

Hobart

Perth

AdelaideBrisbane

Melbourne

Sydney

Canberra

Darwin

0

10

20

30

40

50

60

Billi

on k

ilom

etre

s tr

avel

led

Sydney VKT

Motorcycles

Buses


LCVs

Car

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

0

10

20

30

40

50

60

Billi

on k

ilom

etre

s tr

avel

led

Melbourne VKT

Motorcycles

Buses


LCVs

Car

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

0

5

10

15

20

25

Billi

on k

ilom

etre

s tr

avel

led

Brisbane VKT

Motorcycles

Buses


LCVs

Car

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

64


Figure 2.14 Total projected VKT for Brisbane


Figure 2.15 Total projected VKT for Adelaide


0

10

20

30

40

50

60

Billi

on k

ilom

etre

s tr

avel

led

Melbourne VKT

Motorcycles

Buses


LCVs

Car

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

0

5

10

15

20

25

Billi

on k

ilom

etre

s tr

avel

led

Brisbane VKT

Motorcycles

Buses


LCVs

Car

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

Billi

on k

ilom

etre

s tr

avel

led

Adelaide VKT

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

0

2

4

6

8

10

12

14

16Motorcycles

Buses


LCVs

Car

0

5

10

15

20

25

Billi

on k

ilom

etre

s tr

avel

led

Perth VKT

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

Motorcycles

Buses


LCVs

Car

65


Figure 2.16 Total projected VKT for Perth


Figure 2.17 Total projected VKT for Hobart


Billi

on k

ilom

etre

s tr

avel

led

Adelaide VKT

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

0

2

4

6

8

10

12

14

16Motorcycles

Buses


LCVs

Car

0

5

10

15

20

25

Billi

on k

ilom

etre

s tr

avel

led

Perth VKT

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

Motorcycles

Buses


LCVs

Car

Billi

on k

ilom

etre

s tr

avel

led

Hobart VKT

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0Motorcycles

Buses

Articulated trucks

Rigid trucks

LCVs

Car

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

Billi

on k

ilom

etre

s tr

avel

led

Darwin VKT

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

Motorcycles

Buses


LCVs

Car

66


Figure 2.18 Total projected VKT for Darwin


Figure 2.19 Total projected VKT for Canberra


Billi

on k

ilom

etre

s tr

avel

led

Hobart VKT

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0Motorcycles

Buses

Articulated trucks

Rigid trucks

LCVs

Car

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

Billi

on k

ilom

etre

s tr

avel

led

Darwin VKT

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

Motorcycles

Buses


LCVs

Car

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

Billi

on k

ilom

etre

s tr

avel

led

Canberra VKT

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

Motorcycles

Buses


LCVs

Car

67


Table 2.1 National base case projections of metropolitan vehicle kilometres travelled by type of vehicle, 1990–2020

(billion kilometres)

Fin. Cars LCVs Rigid Articulated Buses Motor Total Year trucks trucks cycles

1990 73.43 10.16 4.102 0.698 0.659 0.856 89.901991 73.84 10.13 3.864 0.648 0.663 0.799 89.941992 75.07 10.40 3.776 0.687 0.666 0.801 91.391993 76.98 10.72 3.726 0.725 0.673 0.819 93.641994 78.56 11.05 3.748 0.768 0.704 0.805 95.631995 81.96 12.14 3.872 0.829 0.733 0.799 100.331996 84.30 12.69 3.863 0.851 0.757 0.752 103.211997 85.21 12.88 3.845 0.918 0.776 0.749 104.391998 86.92 13.48 3.772 0.963 0.798 0.720 106.651999 89.21 14.00 3.797 1.031 0.813 0.693 109.542000 91.24 14.56 3.709 1.032 0.832 0.701 112.072001 91.13 14.56 3.741 1.046 0.861 0.722 112.072002 94.94 15.35 3.879 1.095 0.868 0.729 116.862003 96.37 15.50 3.890 1.106 0.887 0.751 118.502004 101.51 16.29 3.946 1.134 0.906 0.806 124.602005 102.53 16.73 4.089 1.187 0.932 0.872 126.33

2006 103.64 17.72 4.176 1.215 0.949 0.938 128.642007 105.95 18.85 4.311 1.274 0.973 0.989 132.342008 108.54 19.77 4.440 1.342 0.993 1.032 136.122009 111.43 20.63 4.552 1.407 1.009 1.050 140.082010 113.61 21.60 4.652 1.483 1.025 1.069 143.442011 115.34 22.56 4.716 1.552 1.041 1.088 146.302012 116.90 23.53 4.771 1.629 1.056 1.107 148.992013 118.36 24.49 4.814 1.707 1.071 1.125 151.572014 119.79 25.44 4.882 1.784 1.086 1.144 154.132015 121.19 26.40 4.947 1.864 1.105 1.162 156.672016 122.44 27.38 5.025 1.944 1.125 1.181 159.102017 123.67 28.45 5.123 2.036 1.146 1.200 161.622018 124.85 29.54 5.194 2.122 1.167 1.219 164.092019 126.02 30.65 5.294 2.208 1.188 1.238 166.602020 127.30 31.77 5.386 2.292 1.209 1.257 169.21

Growth 2005–2020 24.2% 90.0% 31.7% 93.0% 29.8% 44.2% 33.9%

Note: ‘Metropolitan’ results refer to all activity within the greater metropolitan areas of the eight State and Territory capital cities.

Sources: ABS (2003 and earlier), BTE (1999), BTRE (2002a, 2003a, 2006a), BTRE estimates.

68


Table 2.2 Base case projections of vehicle kilometres travelled by type of vehicle for Sydney, 1990–2020


Fin. Cars LCVs Rigid Articulated Buses Motor Total

Year trucks trucks cycles

1990 23.49 3.34 1.540 0.227 0.218 0.300 29.121991 23.66 3.33 1.451 0.211 0.224 0.280 29.151992 23.87 3.42 1.418 0.224 0.226 0.280 29.441993 24.48 3.52 1.398 0.236 0.229 0.285 30.151994 24.98 3.63 1.405 0.250 0.237 0.280 30.781995 26.06 3.98 1.450 0.269 0.242 0.277 32.271996 26.80 4.15 1.444 0.276 0.246 0.261 33.171997 27.13 4.21 1.437 0.298 0.252 0.260 33.591998 27.45 4.40 1.408 0.312 0.258 0.249 34.081999 28.05 4.57 1.417 0.334 0.265 0.240 34.882000 28.93 4.76 1.384 0.335 0.267 0.242 35.912001 28.90 4.76 1.396 0.339 0.277 0.249 35.922002 30.11 5.02 1.446 0.355 0.271 0.252 37.452003 30.57 5.07 1.450 0.358 0.272 0.259 37.982004 32.22 5.33 1.471 0.368 0.275 0.278 39.932005 32.55 5.47 1.523 0.385 0.283 0.300 40.51

2006 32.92 5.80 1.556 0.394 0.288 0.323 41.282007 33.67 6.17 1.606 0.413 0.296 0.340 42.492008 34.51 6.47 1.655 0.435 0.302 0.355 43.732009 35.44 6.75 1.697 0.457 0.307 0.361 45.022010 36.15 7.07 1.735 0.481 0.312 0.367 46.122011 36.72 7.39 1.759 0.504 0.317 0.373 47.062012 37.24 7.71 1.780 0.529 0.321 0.379 47.952013 37.72 8.03 1.796 0.554 0.326 0.385 48.812014 38.19 8.34 1.822 0.580 0.331 0.391 49.662015 38.66 8.66 1.847 0.606 0.336 0.398 50.512016 39.08 8.98 1.876 0.632 0.343 0.404 51.322017 39.49 9.34 1.913 0.662 0.349 0.410 52.162018 39.89 9.70 1.940 0.690 0.356 0.416 52.992019 40.28 10.07 1.978 0.718 0.362 0.422 53.832020 40.71 10.44 2.013 0.746 0.369 0.428 54.71

Growth 2005–2020 25.1% 90.8% 32.1% 93.6% 30.5% 42.7% 35.0%


69


Table 2.3 Base case projections of vehicle kilometres travelled by type of vehicle for Melbourne, 1990–2020



1990 22.57 2.48 1.178 0.207 0.146 0.196 26.771991 22.62 2.47 1.110 0.192 0.145 0.182 26.721992 22.89 2.53 1.081 0.203 0.144 0.181 27.021993 23.49 2.60 1.063 0.213 0.146 0.184 27.691994 23.99 2.67 1.065 0.225 0.153 0.180 28.281995 25.04 2.91 1.096 0.242 0.160 0.177 29.631996 25.78 3.03 1.089 0.247 0.165 0.166 30.481997 26.06 3.07 1.082 0.267 0.168 0.166 30.821998 26.82 3.22 1.061 0.280 0.175 0.159 31.711999 27.59 3.34 1.068 0.299 0.179 0.153 32.632000 28.09 3.48 1.044 0.300 0.183 0.155 33.252001 28.08 3.48 1.053 0.304 0.190 0.160 33.262002 29.27 3.67 1.092 0.319 0.194 0.161 34.702003 29.67 3.71 1.094 0.322 0.200 0.166 35.162004 31.22 3.89 1.109 0.330 0.205 0.178 36.932005 31.49 4.00 1.149 0.345 0.210 0.193 37.39

2006 31.80 4.24 1.173 0.353 0.213 0.207 37.982007 32.47 4.50 1.210 0.370 0.218 0.218 38.992008 33.23 4.72 1.246 0.390 0.222 0.228 40.042009 34.08 4.93 1.277 0.409 0.225 0.232 41.152010 34.71 5.16 1.305 0.431 0.229 0.236 42.072011 35.20 5.39 1.322 0.450 0.232 0.240 42.832012 35.64 5.62 1.337 0.472 0.235 0.244 43.542013 36.04 5.84 1.349 0.495 0.237 0.248 44.222014 36.44 6.07 1.367 0.517 0.240 0.252 44.882015 36.82 6.30 1.385 0.540 0.244 0.256 45.542016 37.16 6.53 1.406 0.563 0.248 0.260 46.162017 37.49 6.78 1.433 0.590 0.252 0.264 46.802018 37.80 7.04 1.452 0.614 0.256 0.268 47.432019 38.11 7.30 1.480 0.639 0.260 0.272 48.062020 38.45 7.56 1.504 0.663 0.264 0.276 48.72

Growth 2005–2020 22.1% 89.2% 31.0% 92.0% 26.1% 43.3% 30.3%


70


Table 2.4 Base case projections of vehicle kilometres travelled by type of vehicle for Brisbane, 1990–2020



1990 9.04 1.46 0.542 0.101 0.091 0.167 11.401991 9.12 1.46 0.510 0.094 0.093 0.157 11.431992 9.38 1.51 0.503 0.100 0.093 0.159 11.741993 9.59 1.57 0.501 0.107 0.094 0.165 12.031994 9.78 1.63 0.509 0.114 0.100 0.165 12.291995 10.19 1.81 0.531 0.125 0.109 0.165 12.931996 10.47 1.92 0.537 0.130 0.111 0.157 13.321997 10.57 1.95 0.536 0.140 0.113 0.157 13.461998 10.76 2.05 0.527 0.148 0.118 0.152 13.751999 11.21 2.13 0.532 0.159 0.118 0.146 14.302000 11.29 2.22 0.521 0.159 0.124 0.149 14.462001 11.30 2.23 0.526 0.162 0.129 0.154 14.502002 11.81 2.35 0.547 0.170 0.133 0.156 15.172003 12.05 2.38 0.550 0.172 0.137 0.161 15.462004 12.75 2.51 0.559 0.177 0.142 0.174 16.312005 12.93 2.59 0.580 0.186 0.148 0.189 16.62

2006 13.13 2.75 0.594 0.191 0.150 0.204 17.012007 13.48 2.93 0.615 0.200 0.155 0.216 17.592008 13.87 3.08 0.635 0.211 0.159 0.226 18.182009 14.29 3.23 0.653 0.222 0.162 0.231 18.792010 14.63 3.39 0.669 0.235 0.166 0.236 19.322011 14.91 3.55 0.680 0.246 0.169 0.241 19.802012 15.18 3.71 0.689 0.259 0.172 0.246 20.252013 15.43 3.87 0.697 0.272 0.176 0.251 20.692014 15.67 4.03 0.709 0.285 0.179 0.257 21.132015 15.92 4.19 0.720 0.299 0.183 0.262 21.582016 16.14 4.36 0.733 0.312 0.188 0.267 22.002017 16.37 4.54 0.749 0.328 0.192 0.272 22.452018 16.59 4.73 0.761 0.343 0.196 0.278 22.892019 16.80 4.92 0.778 0.357 0.201 0.283 23.342020 17.04 5.11 0.793 0.372 0.206 0.288 23.80

Growth 2005–2020 31.7% 97.4% 36.7% 100.3% 39.1% 52.9% 43.2%


71


Table 2.5 Base case projections of vehicle kilometres travelled by type of vehicle for Adelaide, 1990–2020



1990 6.50 0.85 0.279 0.058 0.066 0.068 7.821991 6.62 0.85 0.263 0.054 0.067 0.064 7.911992 6.69 0.86 0.256 0.057 0.068 0.063 7.991993 6.86 0.88 0.250 0.059 0.067 0.064 8.181994 6.99 0.90 0.250 0.062 0.070 0.063 8.341995 7.29 0.98 0.256 0.067 0.072 0.062 8.731996 7.49 1.02 0.254 0.068 0.074 0.057 8.961997 7.56 1.02 0.250 0.073 0.075 0.057 9.031998 7.93 1.06 0.243 0.076 0.077 0.054 9.451999 8.00 1.09 0.242 0.080 0.078 0.052 9.552000 8.21 1.13 0.235 0.080 0.080 0.052 9.792001 8.15 1.12 0.235 0.080 0.082 0.053 9.722002 8.45 1.18 0.243 0.084 0.083 0.053 10.092003 8.53 1.18 0.242 0.084 0.085 0.055 10.182004 8.94 1.24 0.245 0.086 0.087 0.059 10.652005 8.99 1.27 0.252 0.090 0.088 0.063 10.75

2006 9.04 1.34 0.257 0.091 0.090 0.067 10.892007 9.20 1.42 0.264 0.095 0.091 0.071 11.142008 9.38 1.48 0.271 0.100 0.092 0.074 11.402009 9.59 1.54 0.277 0.105 0.093 0.074 11.672010 9.73 1.61 0.282 0.110 0.094 0.075 11.902011 9.83 1.67 0.284 0.114 0.095 0.076 12.072012 9.92 1.74 0.287 0.120 0.096 0.077 12.242013 10.00 1.80 0.288 0.125 0.097 0.078 12.392014 10.08 1.86 0.291 0.130 0.097 0.079 12.542015 10.15 1.93 0.294 0.135 0.098 0.080 12.692016 10.21 1.99 0.298 0.141 0.099 0.081 12.822017 10.27 2.06 0.302 0.147 0.101 0.082 12.962018 10.32 2.13 0.305 0.153 0.102 0.083 13.102019 10.38 2.21 0.310 0.158 0.103 0.084 13.242020 10.44 2.28 0.315 0.164 0.104 0.085 13.39

Growth 2005–2020 16.1% 79.9% 24.6% 82.6% 18.0% 34.6% 24.5%


72


Table 2.6 Base case projections of vehicle kilometres travelled by type of vehicle for Perth, 1990–2020



1990 7.99 1.45 0.437 0.086 0.081 0.070 10.121991 8.01 1.44 0.412 0.080 0.079 0.065 10.091992 8.24 1.49 0.404 0.085 0.079 0.065 10.361993 8.46 1.54 0.401 0.090 0.081 0.067 10.641994 8.64 1.60 0.405 0.096 0.086 0.066 10.891995 9.02 1.76 0.421 0.104 0.091 0.066 11.461996 9.28 1.85 0.422 0.107 0.095 0.062 11.821997 9.38 1.89 0.422 0.116 0.101 0.062 11.961998 9.46 1.98 0.415 0.122 0.101 0.060 12.141999 9.68 2.06 0.418 0.131 0.104 0.058 12.462000 9.97 2.15 0.410 0.132 0.106 0.059 12.832001 9.99 2.16 0.414 0.134 0.110 0.061 12.862002 10.42 2.27 0.429 0.140 0.113 0.062 13.442003 10.62 2.30 0.431 0.142 0.117 0.064 13.672004 11.21 2.42 0.438 0.145 0.121 0.069 14.402005 11.35 2.49 0.454 0.153 0.124 0.075 14.65

2006 11.51 2.64 0.464 0.156 0.127 0.080 14.982007 11.80 2.81 0.480 0.164 0.131 0.085 15.472008 12.12 2.95 0.495 0.173 0.134 0.089 15.962009 12.47 3.09 0.508 0.182 0.137 0.091 16.472010 12.75 3.24 0.520 0.192 0.139 0.093 16.932011 12.97 3.38 0.527 0.201 0.142 0.095 17.322012 13.18 3.53 0.534 0.211 0.145 0.096 17.702013 13.38 3.68 0.539 0.221 0.147 0.098 18.072014 13.57 3.83 0.548 0.232 0.150 0.100 18.432015 13.76 3.98 0.556 0.242 0.153 0.102 18.802016 13.94 4.13 0.565 0.253 0.156 0.104 19.152017 14.11 4.30 0.577 0.265 0.159 0.106 19.522018 14.28 4.47 0.585 0.277 0.163 0.108 19.882019 14.45 4.64 0.597 0.288 0.166 0.110 20.252020 14.62 4.82 0.608 0.300 0.170 0.112 20.63

Growth 2005–2020 28.8% 93.5% 33.9% 96.3% 36.6% 49.7% 40.8%


73


Table 2.7 Base case projections of vehicle kilometres travelled by type of vehicle for Hobart, 1990–2020



1990 1.13 0.17 0.047 0.009 0.020 0.011 1.381991 1.11 0.16 0.045 0.009 0.019 0.010 1.351992 1.15 0.17 0.043 0.009 0.019 0.010 1.401993 1.18 0.17 0.042 0.009 0.019 0.010 1.431994 1.20 0.18 0.042 0.010 0.020 0.010 1.461995 1.26 0.19 0.044 0.011 0.021 0.010 1.531996 1.29 0.20 0.044 0.011 0.022 0.009 1.581997 1.31 0.20 0.043 0.012 0.022 0.009 1.591998 1.28 0.21 0.042 0.012 0.022 0.008 1.571999 1.35 0.21 0.042 0.013 0.022 0.008 1.642000 1.39 0.21 0.041 0.012 0.023 0.008 1.692001 1.37 0.21 0.041 0.012 0.023 0.008 1.662002 1.40 0.22 0.042 0.013 0.023 0.008 1.712003 1.41 0.22 0.042 0.013 0.024 0.008 1.712004 1.47 0.23 0.043 0.013 0.024 0.009 1.782005 1.47 0.23 0.044 0.013 0.024 0.009 1.79

2006 1.47 0.24 0.045 0.013 0.025 0.010 1.802007 1.48 0.25 0.045 0.014 0.025 0.010 1.832008 1.50 0.26 0.046 0.014 0.025 0.010 1.862009 1.53 0.27 0.046 0.015 0.025 0.011 1.892010 1.54 0.28 0.047 0.015 0.025 0.011 1.912011 1.54 0.28 0.047 0.016 0.026 0.011 1.932012 1.55 0.29 0.047 0.016 0.026 0.011 1.942013 1.55 0.30 0.046 0.017 0.026 0.011 1.952014 1.55 0.31 0.046 0.017 0.026 0.011 1.962015 1.55 0.31 0.046 0.018 0.026 0.011 1.972016 1.55 0.32 0.046 0.018 0.026 0.011 1.972017 1.55 0.33 0.046 0.019 0.026 0.011 1.982018 1.55 0.34 0.046 0.020 0.027 0.011 1.982019 1.54 0.34 0.047 0.020 0.027 0.011 1.992020 1.54 0.35 0.047 0.020 0.027 0.011 2.00

Growth 2005–2020 5.2% 52.3% 5.4% 54.5% 10.6% 16.2% 11.7%


74


Table 2.8 Base case projections of vehicle kilometres travelled by type of vehicle for Darwin, 1990–2020



1990 0.39 0.14 0.025 0.006 0.016 0.009 0.591991 0.42 0.14 0.023 0.006 0.015 0.008 0.611992 0.45 0.15 0.023 0.006 0.015 0.008 0.651993 0.45 0.15 0.022 0.007 0.016 0.009 0.661994 0.46 0.16 0.023 0.007 0.017 0.009 0.671995 0.48 0.17 0.023 0.008 0.017 0.009 0.711996 0.49 0.18 0.024 0.008 0.018 0.008 0.731997 0.50 0.18 0.023 0.008 0.019 0.008 0.741998 0.52 0.19 0.023 0.009 0.020 0.008 0.761999 0.53 0.19 0.023 0.009 0.021 0.008 0.782000 0.55 0.20 0.023 0.009 0.021 0.008 0.812001 0.55 0.20 0.023 0.009 0.022 0.008 0.812002 0.57 0.21 0.024 0.010 0.023 0.008 0.842003 0.59 0.21 0.024 0.010 0.023 0.008 0.862004 0.62 0.22 0.024 0.010 0.024 0.009 0.902005 0.63 0.22 0.025 0.010 0.025 0.010 0.92

2006 0.64 0.23 0.026 0.010 0.026 0.010 0.942007 0.65 0.24 0.026 0.011 0.026 0.011 0.972008 0.67 0.25 0.027 0.011 0.027 0.011 1.002009 0.69 0.26 0.027 0.012 0.028 0.012 1.032010 0.70 0.27 0.028 0.012 0.028 0.012 1.062011 0.72 0.28 0.028 0.013 0.028 0.012 1.082012 0.73 0.29 0.028 0.013 0.029 0.012 1.102013 0.74 0.30 0.028 0.014 0.029 0.012 1.122014 0.75 0.31 0.028 0.014 0.030 0.013 1.152015 0.77 0.32 0.028 0.015 0.030 0.013 1.172016 0.78 0.33 0.028 0.015 0.031 0.013 1.192017 0.79 0.33 0.028 0.016 0.031 0.013 1.212018 0.80 0.34 0.029 0.016 0.032 0.013 1.232019 0.81 0.35 0.029 0.017 0.032 0.014 1.252020 0.82 0.36 0.029 0.017 0.033 0.014 1.28

Growth 2005–2020 30.8% 64.9% 14.2% 67.4% 30.9% 42.9% 39.1%


75


Table 2.9 Base case projections of vehicle kilometres travelled by type of vehicle for Canberra, 1990–2020



1990 2.32 0.27 0.053 0.003 0.021 0.035 2.701991 2.30 0.27 0.050 0.003 0.022 0.033 2.681992 2.41 0.28 0.049 0.003 0.021 0.033 2.801993 2.47 0.29 0.048 0.003 0.021 0.034 2.871994 2.52 0.30 0.049 0.004 0.021 0.033 2.921995 2.63 0.33 0.050 0.004 0.022 0.033 3.061996 2.70 0.34 0.051 0.004 0.025 0.031 3.151997 2.72 0.35 0.052 0.004 0.027 0.030 3.191998 2.70 0.37 0.052 0.005 0.027 0.029 3.191999 2.80 0.39 0.053 0.005 0.026 0.028 3.302000 2.82 0.41 0.053 0.005 0.027 0.028 3.342001 2.80 0.41 0.054 0.005 0.028 0.029 3.322002 2.90 0.43 0.056 0.005 0.028 0.029 3.452003 2.93 0.43 0.057 0.005 0.029 0.030 3.492004 3.10 0.45 0.058 0.006 0.030 0.031 3.672005 3.12 0.46 0.061 0.006 0.030 0.034 3.71

2006 3.14 0.49 0.062 0.006 0.030 0.036 3.762007 3.20 0.52 0.064 0.006 0.031 0.038 3.862008 3.27 0.54 0.065 0.007 0.031 0.039 3.952009 3.34 0.57 0.067 0.007 0.032 0.039 4.052010 3.40 0.59 0.068 0.007 0.032 0.040 4.132011 3.43 0.62 0.069 0.008 0.033 0.040 4.202012 3.47 0.64 0.070 0.008 0.033 0.040 4.262013 3.50 0.67 0.070 0.008 0.033 0.041 4.322014 3.53 0.69 0.071 0.009 0.034 0.041 4.382015 3.56 0.72 0.072 0.009 0.034 0.041 4.432016 3.58 0.74 0.073 0.009 0.035 0.042 4.482017 3.61 0.77 0.074 0.010 0.035 0.042 4.542018 3.63 0.80 0.075 0.010 0.036 0.042 4.592019 3.65 0.83 0.076 0.011 0.036 0.043 4.642020 3.67 0.85 0.078 0.011 0.037 0.043 4.70

Growth 2005–2020 17.9% 84.2% 27.5% 86.9% 23.7% 27.0% 26.6%


76


Table 2.10 National base case projections of total metropolitan vehicle travel–passenger car equivalents, 1990–2020

(billion PCU-km)

Year Syd Mel Bne Adl Per Hob Dar Cbr Total

1990 32.07 29.07 12.53 8.47 11.15 1.51 0.68 2.83 98.31991 31.99 28.92 12.52 8.54 11.08 1.47 0.70 2.81 98.01992 32.29 29.24 12.86 8.63 11.37 1.52 0.73 2.93 99.51993 33.04 29.93 13.17 8.81 11.67 1.55 0.75 3.00 101.91994 33.74 30.57 13.48 8.99 11.96 1.59 0.76 3.06 104.11995 35.41 32.06 14.21 9.42 12.61 1.66 0.80 3.21 109.41996 36.38 32.95 14.65 9.67 13.01 1.71 0.83 3.31 112.51997 36.85 33.34 14.83 9.75 13.18 1.73 0.84 3.35 113.91998 37.40 34.29 15.16 10.17 13.39 1.71 0.87 3.35 116.31999 38.32 35.30 15.76 10.29 13.76 1.78 0.89 3.48 119.62000 39.37 35.93 15.94 10.53 14.14 1.83 0.92 3.52 122.22001 39.41 35.97 15.99 10.47 14.19 1.80 0.92 3.50 122.22002 41.08 37.53 16.74 10.87 14.83 1.86 0.96 3.64 127.52003 41.63 38.00 17.04 10.96 15.07 1.86 0.98 3.68 129.22004 43.68 39.86 17.94 11.45 15.85 1.93 1.02 3.87 135.62005 44.38 40.41 18.31 11.57 16.15 1.94 1.04 3.91 137.7

2006 45.28 41.10 18.76 11.73 16.53 1.95 1.07 3.97 140.42007 46.67 42.25 19.43 12.02 17.10 1.99 1.10 4.08 144.62008 48.08 43.44 20.10 12.32 17.66 2.02 1.14 4.18 148.92009 49.53 44.67 20.78 12.62 18.24 2.06 1.17 4.29 153.42010 50.81 45.72 21.40 12.88 18.77 2.08 1.20 4.38 157.22011 51.91 46.60 21.95 13.08 19.23 2.10 1.23 4.45 160.62012 52.96 47.44 22.49 13.28 19.68 2.11 1.25 4.52 163.72013 53.96 48.23 23.01 13.46 20.11 2.13 1.28 4.59 166.82014 54.98 49.02 23.53 13.64 20.55 2.14 1.30 4.65 169.82015 55.99 49.81 24.05 13.82 20.98 2.15 1.33 4.72 172.82016 56.97 50.56 24.57 13.99 21.41 2.16 1.35 4.77 175.82017 58.01 51.35 25.11 14.16 21.86 2.17 1.38 4.84 178.92018 59.02 52.11 25.64 14.34 22.30 2.18 1.41 4.90 181.92019 60.06 52.89 26.19 14.51 22.75 2.19 1.43 4.96 185.02020 61.12 53.70 26.75 14.69 23.21 2.20 1.46 5.03 188.2

Growth 2005–2020 37.7% 32.9% 46.1% 27.0% 43.8% 13.4% 40.2% 28.5% 36.6%

Note: PCU-km estimates calculated using traffic contribution values (relative to passenger car = 1) of: LCV = weighted sum of standard light commercials (PCU=1) and large vans (PCU=1.5); Rigid truck = 2; Articulated truck = weighted sum of standard 6-axle semi-trailer (PCU=3) and B-doubles (PCU=4); Motorcycle = 0.5; and Bus = 2.

Source: BTRE (2003a, 2006a), BTRE estimates.

77


The costs of congestion

Congestion imposes significant social costs with interruptions to traffic flow lengthening average journey times, making trip travel times more variable and making vehicle engine operation less efficient. The cost estimates presented here include allowances for:

• extra travel time (e.g. above what would have been incurred had the vehicle been travelling under more freely flowing conditions),

• extra travel time variability (where congestion can result in trip times becoming more uncertain, leading to travellers having to allow for an even greater amount of travel time than the average journey time, in order to avoid being late at their destination),

• increased vehicle operating costs (primarily higher rates of fuel consumption), and

• poorer air quality (with vehicles under congested conditions emitting higher rates of noxious pollutants than under free-flow conditions, leading to even higher health costs).

Most studies that give congestion cost estimates are not entirely suitable for assessing the real impacts of congestion on society—or for valuing the avoidable social costs of congestion (i.e. those costs that could be saved under appropriate policy or operational intervention). This is because they are usually based on the total time costs for congestion delay—which refer to the differences in costs (borne by society) between average travel at the current congested traffic levels and travel under totally uncongested conditions. However, some level of delay is practically unavoidable if there are a reasonable number of interacting vehicles on a particular road link—that is, under real-world traffic conditions, it is typically not practicable (nor desirable on social or economic efficiency grounds) to reduce congestion to zero. To enable a particular vehicle to travel at free-flow speeds during a time of peak underlying demand would essentially require the suppression of most of that demand, and consequent high social costs to those not able to travel then.

‘Total cost of congestion delay’ estimates are derived by the BTRE methodology but only as intermediate values. Since total delay values

78


are based on the value of the excess travel time compared with travel under completely free-flow conditions—an unrealisable situation for actual road networks—they are rather poor measures of the social gains that could be obtained through actual congestion reduction, and are only given in this paper for comparison purposes. A preferable set of values to use for the ‘social costs of congestion’ are the estimated deadweight losses (DWLs) associated with a particular congestion level, which basically gives a measure of the cost of doing nothing about current congestion or the avoidable costs of traffic congestion levels.

That is, DWL valuations give an estimate of how much total costs (for time lost and other wasted resources) could be reduced if traffic volumes were reduced to the economically optimal level (which is defined as the level of traffic beyond which the full social costs of any further travel would outweigh the benefits of that extra travel). For most urban travel, including in times of peak demand, motorists only make an allowance for the time costs they are likely to face personally (i.e. the current average generalised cost) and do not also take into account all the extra delay that their entry into the already congested traffic stream is likely to impose on other motorists (i.e. allow for the marginal generalised cost for current traffic levels). If a pricing mechanism could be put in place that obliged all motorists to base their travel decisions not only on their private costs (represented by the current average cost values) but also on the additional costs imposed on others (the external costs—represented by the difference between the average and marginal costs), then transport choices would alter. Some trips would be deferred or not taken, others would be re-routed or taken on alternative modes. The resulting traffic flow (commonly termed the socially or economically optimal quantity of travel) would involve a somewhat lower vehicle density than before—but nowhere near as low as that required for free-flow speeds—and would theoretically have significantly lower congestion delay.

This BTRE study first estimates total delay costs, and then derives the DWL portion of those costs—and gives the final results for estimated social costs of congestion in terms of the deadweight losses associated with a respective congestion intensity and its level of total delay. (DWLs appear to be in the order of half total delay costs for typical peak traffic conditions—where their proportion would be much lower for light traffic and grow rapidly for severely congested areas).

79


The calculation of generalised costs requires the conversion of estimates of hours lost due to congestion into dollars—that is, a unit value of time. The unit cost rates used in the BTRE congestion estimation process (described in the following sections) were derived from standard Austroads values for road user costs (Economic Evaluation of Road Investment Proposals—Unit Values for Road User Costs at June 2002, Austroads 2004). The Austroads report gives different estimated values of travel time for different vehicle types, with private vehicle travel incurring costs in order of $9 per person-hour and business travel in the order of $20–$30 per person-hour. Freight vehicles get a further cost rate in terms of dollars per vehicle-hour (e.g. with a 6-axle articulated truck having an additional travel value of around $28 per vehicle-hour).

The BTRE analyses also include estimates of health/damage costs for urban air pollution, requiring the use of unit costs for motor vehicle emissions (e.g. in terms of dollars per tonne emitted). The unit cost rates used in this BTRE study are based on estimates of the health impacts of Australian air pollution by Paul Watkiss (Fuel Taxation Inquiry: The Air Pollution Costs of Transport in Australia). The Watkiss (2002) values for the costs of environmental damage due to air pollution vary greatly between the different emission species, ranging from $3 per tonne for carbon monoxide up to high costs of $342 000 dollars per tonne for particulate matter (for inner-city areas of the larger capitals). These unit costs imply that urban traffic contributes, on average, around 3.6 cents per vehicle kilometre travelled to the total social costs of air pollution (with about 2.5 c/km for cars and over 20 c/km for heavy vehicles).

The Bureau has also recently published a study on the costs of air pollution (BTRE 2005, Health Impacts of Transport Emissions in Australia: Economic Costs, Working Paper 63). The total (mortality and morbidity) cost values derived in Working Paper 63 are of a comparable magnitude to the Watkiss damage valuations, with metropolitan totals using Watkiss unit costs coming out higher than Working Paper 63’s ‘central’ estimate for motor vehicle air pollution costs, but within the uncertainty/sensitivity range given around this central estimate. Unit values derived from the Working Paper 63 total costs were not suitable for this current congestion study—since that report ‘estimated the long-term health impacts of ambient air pollution using particulate matter of less than 10 microns as a surrogate for all air pollutants’. The

80


contribution of congestion to total emission costs requires the use of speed versus emission rate curves. And since the different emission species have reasonably different engine load versus emission response relationships, separate unit costs are needed for each individual emission type included in the analysis. This meant that the Watkiss (speciated) unit values were more useful for this study than a solely PM10 unit rate derived from Working Paper 63 costings.

BTRE modelling (e.g. see BTCE 1996b, BTRE 2003a) implies that traffic interruptions due to road congestion account for around 15 to 35 per cent of the emissions generated by urban motor vehicles, depending on the emission species, by increasing emission rates to higher than average levels during interrupted travel conditions. With congestion growing, this percentage is projected to increase by 2020. This current study attributes a congestion share of estimated total health costs from motor vehicle pollution according to calculated percentage ratios (for the proportional contribution of various levels of congestion, based on the various speed-emission rate curves).

An important point to stress is that the current BTRE approach is basically an aggregate modelling one—that is, it does not use detailed network models (which separately model traffic flows on all the cities’ major roads), but aims to roughly estimate the scale of a city’s congestion situation using aggregate indicators of a city’s overall average traffic conditions. The main advantages of this aggregate approach relate to the ability to generate congestion cost estimates and projections with relatively slight computational and data resources. (Network models tend to require extensive data input and a considerable level of computational and maintenance support.) The main disadvantages relate to congestion being such a non-linear, inhomogeneous (in fact, so location-specific that certain bottlenecks can account for an inordinate amount of an area’s total delay) and stochastic process that highly accurate assessments of its impacts can really only be accomplished using detailed network models.

The following congestion cost estimates are therefore provided as ‘order of magnitude’ evaluations—to help with considerations dealing with the likely aggregate costs of urban transport externalities, and their likely future trends. Detailed assessments of each of the cities’ congestion impacts, especially analyses at the level of particular major city roads or thoroughfares, should be pursued using the appropriate

81


network modelling frameworks and the Bureau firmly encourages the potential of various jurisdictions undertaking more congestion modelling on the network (and whole of day) scale, particularly using micro-simulation techniques. (Especially now that more powerful and less computationally resource hungry micro-simulation models are emerging and becoming more prevalent in the transport modelling arena, though it should be noted that wide-scale micro-simulation models are still highly data and calibration intensive responsibilities, e.g. see Austroads 2006a). Conducting such modelling on consistent bases will also allow more comparisons to be made between studies done in the various cities.

Furthermore, additional (network modelling) work dealing with the separation of total congestion delay into recurrent and incident-based components would also be useful. Incidents, such as road accidents and bad weather, contribute to a significant element of total delay; and their analysis is often best handled using traffic microsimulation approaches.

BTRE congestion work has typically used speed-flow relationships of the form:

ATT = T (1+a{(x-1)+((x-1)2 + bx)0.5})

where

ATT = Average travel time per kilometre,

T = Free speed travel time per kilometre,

x = volume-capacity ratio, and

a and b are adjustable parameters;

following the derivations of Kimber and Holis (1979) and Akcelik (1978, 1991).

Figure 1.20 plots some examples of typical speed-flow curves for a variety of road types, and an area averaged curve that results from aggregating delay over a set of road links and intersections (derived using network models). Though the current analysis does not entail running network models, the methodology does rely on previous Bureau results using network models for Australian metropolitan traffic assessment (see BTCE Report 92 for some details of analyses

82


using the TRANSTEP trip assignment model). Part of the Bureau network modelling work allowed the semi-empirical derivation of aggregate speed-flow response curves, by averaging over sets of links and intersections within a defined area.

To derive the current BTRE congestion cost estimates, a set of aggregate speed-flow curves were used for the analysis, using aggregate data on traffic flows for the Australian capital cities (i.e. relating average vehicle travel speeds to the average road network volume-to-capacity ratios for the major road types and areas of the city). The functional forms were partly calibrated using data on average speeds over the cities’ arterial and major road networks (from the Austroads National Performance Indicators) and BTRE estimates of urban VKT by the various vehicle types (as shown in figures 2.3 to 2.19). The BTRE base case projections (of city-by-city VKT growth, along with assumptions about likely trends in road capacity growth) then allow estimates of likely average speed reductions for future traffic levels.

A sample of some of the vehicle speed response curves used within the BTRE models, for fuel consumption and vehicle emission rates, are given in figures 2.21 and 2.22.

Figure 2.20 Representative functional forms for normalised speed-flow relationships for various road types

0.0

0.2

0.4

0.6

0.8

1.0Uniterrupted freeway or arterial

Freeway with interruptions

Average for dense urban network

arterial road with interruptions

Secondary road

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6

Inde

x of

spe

ed (r

elat

ive

to fr

ee-fl

ow c

ondi

tions

)

Increasing volume-capacity ratio for road

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5 Petrol carDiesel LCV

Articulated truck

Inde

x of

fuel

con

sum

ptio

n re

lativ

e to

40k

m/h

r

Vehicle speed (km/hr)

0 10 20 30 40 50 60 70 80 90 100 110 120

83


Figure 2.21 Typical response curves for average fuel consumption to variation in vehicle speed

Figure 2.22 Typical response curves for petrol car emission rates to variation in speed, by pollutant type

0.0

0.2

0.4

0.6

0.8

1.0Uniterrupted freeway or arterial

Freeway with interruptions

Average for dense urban network

arterial road with interruptions

Secondary road

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6

Inde

x of

spe

ed (r

elat

ive

to fr

ee-fl

ow c

ondi

tions

)

Increasing volume-capacity ratio for road

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5 Petrol carDiesel LCV

Articulated truck

Inde

x of

fuel

con

sum

ptio

n re

lativ

e to

40k

m/h

r


0 10 20 30 40 50 60 70 80 90 100 110 120

% o

f dai

ly tr

affic

Hour of day

0

1

2

3

4

5

6

7

8

9

Network (arterial) average

Freeway average

0 1 2 6543 12 13 14 15 16 17 18 19 20 21 22 23 241110987

Representative daily traffic profile–total flows

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0Fuel consumption

NOx emissions

VOC emissions

PM emissions

CO emissions

Inde

x of

g/k

m re

lativ

e to

40

km/h

r


0 10 20 30 40 50 60 70 80 90 100

84


The following section outlines the current BTRE methodology, giving a primarily graphical overview of the steps involved in the congestion cost estimation process, using the various functional forms derived (relating aspects of congestion and overall network performance to average levels of traffic using the network at various times of the day). For a schematic diagram briefly summarising the cost estimation process see Appendix figure A.3.

Total congestion cost estimation

BTRE estimates of total metropolitan VKT for a particular city (e.g. as shown in figures 2.12 to 2.19) are first used to derive an average daily traffic level for the entire network, which is then subdivided into VKT on each of the main road types for each hour of the day, using average daily traffic profiles for the current city networks.

Figure 2.23 Hourly traffic volumes for typical metropolitan travel

The hourly VKT levels are then subdivided between the various vehicle types (according to the total vehicle-specific utilisations given in figures 2.12 to 2.19) using average daily profiles derived for the major components of overall traffic.

% o

f dai

ly tr

affic

Hour of day

0

1

2

3

4

5

6

7

8

9

Network (arterial) average

Freeway average

0 1 2 6543 12 13 14 15 16 17 18 19 20 21 22 23 241110987

Representative daily traffic profile–total flows

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0Fuel consumption

NOx emissions

VOC emissions

PM emissions

CO emissions

Inde

x of

g/k

m re

lativ

e to

40

km/h

r


0 10 20 30 40 50 60 70 80 90 100

85


For example, the hourly pattern of private car use is very close to the average profile shown above, while that for freight vehicles tends to involve a greater proportion of travel outside of the traditional peak periods (where a representative profile for trucks is given in Figure 2.24).

Figure 2.24 Typical daily traffic profile for commercial vehicles

Using the various road type and vehicle-specific daily profile curves gives the following (figures 2.25a and 2.25b) general shapes for hourly distributions of average Australian metropolitan traffic throughout the day (for a typical weekday).

Figure 2.25a Typical daily VKT profile by vehicle type

VKT

inde

x

Hour of day

0 1 2 6543 12 13 14 15 16 17 18 19 20 21 22 23 241110987

Typical composition of total VKT by road class

0.0

0.5

1.0

1.5

2.0

2.5

3.0

CBD

Freeway/highway

Arterials

Local roads

0

1

2

3

4

5

6

7

8

Hour of day

0 1 2 6543 12 13 14 15 16 17 18 19 20 21 22 23 241110987

% o

f dai

ly tr

affic

Representative daily profile–heavy vehicles (network average)

Motorcycles

Buses

Articulated trucks

Rigid trucks

LCVs

Cars

0

1

2

3

4

5

6

7

8

9

Hourly VKT by vehicle type

Hour of day

0 1 2 6543 12 13 14 15 16 17 18 19 20 21 22 23 241110987

% o

f tot

al d

aily

VKT

86


The application of average vehicle occupancy rates to these VKT profiles then allows the pattern of daily passenger-kilometres to be estimated (with the occupancies also varying by time of day, e.g. average bus loading levels being much higher during peak periods than during the off-peak).

Note: for estimating passenger tasks of LCVs and trucks (i.e. as opposed to freight or service uses), the non-business portion of total VKT is used.

Figure 2.25b Typical daily VKT profile by road type

Applying occupancy rates to the VKT pattern plotted in figures 2.25a and 2.25b, after allowing for the variation of average vehicle occupancy rates over the hours of the day, results in the following estimated profile for metropolitan passenger task (pkm) by time of day.

VKT

inde

x

Hour of day

0 1 2 6543 12 13 14 15 16 17 18 19 20 21 22 23 241110987

Typical composition of total VKT by road class

0.0

0.5

1.0

1.5

2.0

2.5

3.0

CBD

Freeway/highway

Arterials

Local roads

0

1

2

3

4

5

6

7

8

Hour of day

0 1 2 6543 12 13 14 15 16 17 18 19 20 21 22 23 241110987

% o

f dai

ly tr

affic

Representative daily profile–heavy vehicles (network average)

87


Figure 2.26 Typical daily passenger task profile by vehicle type

Using the appropriate aggregate speed-flow curves for the various road types and city areas, and summing over the resulting estimates of average travel speeds (weighted by the VKT distributions derived above) for each hour of the day then gives an estimate for the network-wide average travel time pattern over the day.

Figure 2.27 Average urban traffic performance by time of day

% o

f tot

al d

aily

pkm

Hour of day

0 1 2 6543 12 13 14 15 16 17 18 19 20 21 22 23 241110987

Proportion of total passenger–kilometres by vehicle type

Motorcycles

Buses

LCVs

Cars

0

1

2

3

4

5

6

7

8

9

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

Hour of day

0 1 2 6543 12 13 14 15 16 17 18 19 20 21 22 23 241110987

Min

/km

(ave

rage

trav

el)

Average network travel time

% o

f tot

al d

aily

pkm

Hour of day

0 1 2 6543 12 13 14 15 16 17 18 19 20 21 22 23 241110987

Proportion of total passenger–kilometres by vehicle type

Motorcycles

Buses

LCVs

Cars

0

1

2

3

4

5

6

7

8

9

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

Hour of day

0 1 2 6543 12 13 14 15 16 17 18 19 20 21 22 23 241110987

Min

/km

(ave

rage

trav

el)

Average network travel time

88


Using estimated or assumed ‘free’ speeds (i.e. average traffic speeds encountered under totally uncongested conditions) for each major road type then allows average delays (free speed minus actual traffic speed) to be calculated. A significant part of the uncertainty associated with this type of congestion estimation methodology relates to the setting of the free speed values, since they are typically not precisely known for most real-world travel. The values used in the current analyses are a combination of floating-car values for off-peak travel (when available), estimates of likely free speed levels based on the measured differences between peak speeds and traffic volumes versus inter-peak speeds and volumes (when fully off-peak values not available), estimated network free speeds from the Bureau TRANSTEP modelling work, and literature values for various cities (which are typically derived from network models, calibrated to floating-car records, or traffic monitoring data).

Several cities report annual averages for their posted speed limits (i.e. traffic weighted averages across the whole city’s major road network, including all arterials and freeways, of the speed limits on each road link) to Austroads’ National Performance Indicators. These values tend to vary somewhat from city to city, and typically fall in the range of between 65 to 75 kilometres per hour. Due to road lay-out and design factors (such as the density and control of intersections), actual traffic on these city networks will not generally be able to travel at the posted speed limits even during uncongested traffic conditions. The current BTRE model specification has input values for approximate network-wide free speeds typically ranging between about 56 to 65 kilometres per hour.

The resulting pattern for average delay over the day (for an average week-day), summed over the road types, for a typical major metropolitan area is shown in the next graph (figure 2.28).

89


Figure 2.28 Typical delay profile

Note: Total delay refers to the comparison of trip times at average recorded traffic speeds and at free-flow speeds for the network.

The total delay is spread over the main vehicle types as shown in Figure 2.29 (estimated using the vehicle-specific daily profiles discussed previously).

Figure 2.29 Typical daily delay profile by vehicle type

% o

f hou

rs (t

otal

del

ay)

Hour of day

Proportion of total daily delay by vehicle type

0

2

4

6

8

10

12

14

16

Hour of day

% o

f dai

ly tr

affic

del

ay

Proportion of total daily delay for average city traffic

0

2

4

6

8

10

12

14

Motorcycles

Buses

Articulated trucks

Rigid trucks

LCVs

Car

0 1 2 6543 12 13 14 15 16 17 18 19 20 21 22 23 241110987

60 1 2 543 12 13 14 15 16 17 18 19 20 21 22 23 241110987

% o

f hou

rs (t

otal

del

ay)

Hour of day

Proportion of total daily delay by vehicle type

0

2

4

6

8

10

12

14

16

Hour of day

% o

f dai

ly tr

affic

del

ay

Proportion of total daily delay for average city traffic

0

2

4

6

8

10

12

14

Motorcycles

Buses

Articulated trucks

Rigid trucks

LCVs

Car

0 1 2 6543 12 13 14 15 16 17 18 19 20 21 22 23 241110987

60 1 2 543 12 13 14 15 16 17 18 19 20 21 22 23 241110987

90


Some literature values (and a variety of congestion indicators often used by road authorities) are based on delay values relative to ‘nominal’ speeds (i.e. based on average posted speed limits) rather than comparisons to estimated ‘free’ speeds. This practice has the advantage of nominal speeds being more precisely defined than free speeds, and tending to be much easier to estimate than free speed values. However, such comparisons to nominal speeds are not as useful for congestion cost estimation since travel at nominal speeds is often not possible for many actual road links, even in times of zero congestion. Free-flow speeds for many roads are typically dependent on the particular road design, especially its number of intersections (and other signalised or unsignalised impedances), and therefore, realistic free speeds are often (particularly for inner-city streets) considerably below posted or nominal speeds.

For example, if the previous delay calculations (as shown in figure 2.28) were done using average nominal speeds, as opposed to estimated free speeds, the total daily delay calculated would be significantly higher, simply due to the chosen comparison speed for defining ‘delay’. The typical order of magnitude for this definitional divergence is illustrated in the next graph (figure 2.30).

Figure 2.30 Divergence in estimated delay depending on comparison speeds

% o

f tot

al c

osts

Hour of day

Typical components of total daily delay costs

0

2

4

6

8

10

12

14

Buses

Motorcycles and private car useBusiness car use

LCVs

Trucks

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

Inde

x of

tota

l hou

rs o

f del

ay

Differences in calculated delay

Houly delay relative to ‘nominal’ speeds

Hourly delay relative to estimated ‘free’ speeds

Hour of day

0 1 2 6543 12 13 14 15 16 17 18 19 20 21 22 23 241110987

0 1 2 6543 12 13 14 15 16 17 18 19 20 21 22 23 241110987

91


Using the previously discussed unit values for travel time (based on the Austroads standards) then gives estimates for the dollar values of total delay (i.e. travel time lost, relative to free speeds) due to congestion. The estimates assume that since around 10 per cent of urban journeys are under 2–3 kilometres in length, that a portion of total trips will not tend to take long enough on average to incur noticeable delay. The current model setting is for an assumed five per cent of trips to be below the threshold of incurring noticeable delay–and are allocated zero delay costs. It is also assumed that average delays incurred on local roads are considerably below those encountered on major metropolitan arterials and freeways (especially during peak periods).

Another adjustable parameter in the model relates to the proportion of trips that (due to particular trip purposes, origin-destinations or individual user preferences) are likely to be less time-sensitive than average. The current assumption (input to the base case scenario) has around 10 per cent of light vehicle travel and around 5 per cent of heavy vehicle travel as being less time-sensitive trips (with the parameter value varying throughout the day, such that time insensitivity is more likely outside of business hours than during them), which are allocated a unit time delay cost of half the standard rate. All these parameters are adjustable, as better data on trip distributions and travel time valuations become available.

Note that the congestion cost values derived by this study, even though given in dollar terms, are not directly comparable to aggregate income measures (such as GDP). Some elements involved in the costings have GDP implications (e.g. the timeliness and reliability of freight and service deliveries will impact on business productivity levels). However, a major proportion of the derived cost values refer to elements that play no part in the evaluation of GDP, such as private travel costs.

Any reduction in congestion delay due to some traffic management measure will typically have some time savings benefits for road users; but the size of those benefits (in dollar terms, using generalised costings)—especially with regard to private individuals—will not give a direct estimate of the size of any GDP changes that happen to flow from the congestion reduction.

92


Total delay cost values

BTRE preliminary analyses give an Australian total of $11.1 billion for total (i.e. relative to free-flow) annual delay costs over the Australian capitals for 2005 (comprising $5.7 billion in private vehicle delay and $5.4 billion in business vehicle delay)—with Sydney (at $3.9 billion), Melbourne (at $3.6 billion) and Brisbane (at $1.44 billion) comprising a major portion of this total. The other cities contributed $0.78 billion for Adelaide, $1.05 billion for Perth, $80 million for Hobart, $27 million for Darwin, and $182 million for Canberra.

The average distribution of these costs over the day, by vehicle type, is displayed in the following graph (figure 2.31).

Figure 2.31 Distribution of total delay costs by vehicle type

The base case demand projections (coupled with the speed-flow and daily profile functions in the BTRE models) have this value of total metropolitan delay rising to $23 billion by 2020 (with private travel incurring $10.9 billion and business vehicle use $12.1 billion). The city specific levels rise to $8.3 billion for Sydney, $7.0 billion for Melbourne, $3.4 billion for Brisbane, $1.4 billion for Adelaide, $2.4 billion for Perth, $0.11 billion for Hobart, $50 million for Darwin, and $0.3 billion for Canberra.

% o

f tot

al c

osts

Hour of day

Typical components of total daily delay costs

0

2

4

6

8

10

12

14

Buses

Motorcycles and private car useBusiness car use

LCVs

Trucks

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

Inde

x of

tota

l hou

rs o

f del

ay

Differences in calculated delay

Houly delay relative to ‘nominal’ speeds

Hourly delay relative to estimated ‘free’ speeds

Hour of day

0 1 2 6543 12 13 14 15 16 17 18 19 20 21 22 23 241110987

0 1 2 6543 12 13 14 15 16 17 18 19 20 21 22 23 241110987

93


To these total delay costs, the methodology then adds an (additional) allowance for trip variability (as a measure of system reliability). Trip variability in uncongested traffic conditions tends to be quite low, but variability in travel time grows as the traffic level increases toward the road’s rated capacity. Over a fairly congested road link, the travel times will tend to exhibit quite a large spread, with the maximum time taken to cover the link often being several times larger than the minimum time.

The current BTRE variability assessment uses measures of the spread of travel times over different trips for a certain time of the day–based on Austroads National Performance Indicators data for average network travel variability and on analyses of the slopes of the derived travel time functional forms (described above, with regards to figure 2.20). The extra travel time required by a certain level of speed variability then has dollar values attached to it, again using the Austroads unit values of time. With respect to the unit values for delay (e.g. in terms of minutes lost per km), an elasticity of 0.8 for private travel and of 1.2 for business travel (where many businesses tend to value trip reliability very highly for product deliveries and service distribution) is applied to one standard deviation (SD, in min/km) around the mean travel time for the relevant period.

These expanded time costs thus make some allowance for:

• the extra time lost due to travellers having to leave earlier than they would otherwise choose (if they could rely on average trip times being fairly constant) to avoid being late at their destinations; and

• the costs to individuals and businesses of not only having their average travel take substantially longer in peak traffic conditions, but that trip times are also more likely to be unpredictable, and that they will often fail to meet schedules.

A representative functional form (for the dependence of trip variability on increasing traffic levels), derived from the average trip time functions presented earlier, is shown in figure 2.32. The functional expressions for such curves (e.g. see Ensor 2002) are typically of the form:

94


Var = S0 + S1/(1 + exp(-c(x - d)))

where

Var = the standard deviation in average trip times (mins/km); x = volume-capacity ratio of the road’s traffic level; S0, S1, c and d are adjustable (positive) parameters depending on the road type.

Figure 2.32 Representative variation of trip reliability with traffic volume

Each period’s average trip variability value—in terms of the extra minutes taken per kilometre that slower travel (at one standard deviation above the mean travel time for that period) entails—is calculated from the relevant traffic volume estimates. The difference is then taken between estimated free-flow trip variability and the estimated trip variability values for the various time periods (and vehicle types). These differences are multiplied by the VKT for that period, summed over the hours of the day, and multiplied by the relevant vehicular value of time, thus giving an estimate for the total cost of trip variability due to congestion.

This method of valuing trip variability may tend to somewhat underestimate the total actual costs to businesses since it does

Business delay cost

Total trip variability

Private delay cost

Inde

x of

trav

el ti

me

spre

ad (1

SD

of a

vera

ge tr

ip ti

mes

)

Typical functional form for travel time variability

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.05

0.15

0.25

0.35

0.45

0.55

0.65

0.75

0.85

0.95

1.05

1.15

1.25

1.35

1.45

1.55

1.65

Average V/C ratio

0

2

4

6

8

10

12

14

% o

f tot

al d

aily

cos

ts

Hour of day

Components of total time costs

0 1 2 6543 12 13 14 15 16 17 18 19 20 21 22 23 241110987

95


not allow for wider economic effects than those directly related to congestion delay for their operating vehicles. Some other impacts congestion can have on businesses include possible reductions in market accessibility; restrictions on locations, inventory practices and specialised labour (or input materials) reducing possible economies of scale; and possibly increased labour costs (associated with wage rates tended to having to compensate workers for their higher commuting costs). Assessing such costs is beyond the scope of this study though they are likely to be substantial in magnitude. Some studies suggest that these broader economic costs to business could be comparable in size to the direct travel costs due to congestion (and, based on this, a rough sensitivity result for their possible impact on the total estimated costs is derived in a following section of the report).

The chosen parameters for the variability costings could be considered conservative–since they are based on a single standard deviation in travel time (i.e. covering variation in travel times to about the 68th percentile, if normally distributed) and an elasticity of approximately one (for the value of time lost due to trip variability versus the value of time lost due to standard delay). However, the engineering definitions of trip variability are commonly based upon 85th percentile travel times (i.e. approximately 1.44 standard deviations from the mean), and some literature values imply that surveyed travellers often express a significantly higher value of time for trip variability losses, as opposed to time losses from expected levels of (recurrent) delay.

This current method of valuing trip variability adds around 25 per cent to the total delay costs. For example, the total time cost for Australian metropolitan areas—i.e. the value of total hours of delay (due to average travel times being above free flow times) plus the extra hours for trip variability (above free-flow levels of variability)—for 2005 has been estimated by our model to be around $14.1 billion dollars. The base case projections push this to around $30 billion for 2020.

The average distribution typical of current total time costs (due to congestion-related delay and travel time variability) over the hours of the day is displayed in figure 2.33.

96


Figure 2.33 Typical daily profile of total time costs

Notes: Costs here refer to on-road travel times compared with free-flow conditions. Costs calculated using VKT data for a base year of 2000.

When projecting the congestion costs into the future, the BTRE models allow for some travel time substitution effects. If increasing traffic levels lead to a period’s generalised travel costs rising sharply enough (as will tend to happen in peak periods where travel times are already much longer than average, and where the marginal costs of travel are particularly high), then a portion of the increasing VKT has its trip timing decision deferred or brought forward, into a period of lesser generalised cost. This leads to a certain amount of peak-spreading in the projected daily profiles, with the amount of traffic in the shoulder periods (around the main peak times) and in the inter-peak period tending to increase by a greater percentage than for the peak volumes.

Figure 2.34 displays how the shape of the daily VKT profile alters over the term of the projection period (for average metropolitan travel in the base case), where even though there is still significant growth in the peak periods, there is substantially greater proportional growth in the hours around the standard peak hours.

Business delay cost


Private delay cost

Inde

x of

trav

el ti

me

spre

ad (1

SD

of a

vera

ge tr

ip ti

mes

)

Typical functional form for travel time variability

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.05

0.15

0.25

0.35

0.45

0.55

0.65

0.75

0.85

0.95

1.05

1.15

1.25

1.35

1.45

1.55

1.65

Average V/C ratio

0

2

4

6

8

10

12

14

% o

f tot

al d

aily

cos

ts

Hour of day

Components of total time costs

0 1 2 6543 12 13 14 15 16 17 18 19 20 21 22 23 241110987

97


Figure 2.34 Projected daily VKT profile

This pattern of limited growth in peak periods, while growth in periods around the peak remains strong, is already apparent in recent yearly data for particular city links (due to many major metropolitan roads already operating close to their rated capacity at certain times of day). For example, the following growth patterns (figure 2.35, reproduced from VicRoads Traffic Systems Performance Monitoring Information Bulletin) have been reported for Melbourne’s freeway traffic over the last few years–where practically all the growth in traffic volume has occurred outside of the times of highest total volume (i.e. the peaks at 8 am and 5:30 pm).

Representative change in daily VKT profile

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

Hour of day

Year 2020 traffic

Year 2000 traffic

Tota

l VKT

inde

x

200

150

100

50

0

-50

Freeways 2005

Freeways 2004

Freeways 2003

Freeways 2002

Freeways 2001

Abso

lute

cha

nge

in a

vera

ge v

olum

e pe

r site

re

lativ

e to

bas

e ye

ar 2

001

for e

ach

15 m

inut

e in

terv

al

Absolute change in 15 minute volumes per site relative to base year 2001

0 1 2 6543 12 13 14 15 16 17 18 19 20 21 22 23 241110987

0 1 2 6543 12 13 14 15 16 17 18 19 20 21 22 231110987

98


Figure 2.35 Growth in Melbourne freeway volumes

Source: VicRoads Information Bulletin, Traffic Systems Performance Monitoring 2004/2005

The projected peak spreading leads to the forecast daily profile for total time costs (associated with congestion) to flatten somewhat, over the course of the day, by 2020. For example, if the shape of the daily cost profile curve representative of year 2000 average traffic volumes (presented in figure 2.33) is compared to the 2020 projected curve (given in figure 2.36), then it is apparent that the contributions of the peaks are much less prominent by 2020.

Figure 2.36 Projected daily profile of total time cost

Note: Costs here refer to travel compared to free-flow conditions

Business delay cost


Private delay cost

0

2

4

6

8

10

12

14

% o

f tot

al d

aily

cos

ts

Hour of day

Components of total time costs: 2020 projections

0 1 2 6543 12 13 14 15 16 17 18 19 20 21 22 23 241110987

Representative change in daily VKT profile

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

Hour of day

Year 2020 traffic

Year 2000 traffic

Tota

l VKT

inde

x

200

150

100

50

0

-50

Freeways 2005

Freeways 2004

Freeways 2003

Freeways 2002

Freeways 2001

Abso

lute

cha

nge

in a

vera

ge v

olum

e pe

r site

re

lativ

e to

bas

e ye

ar 2

001

for e

ach

15 m

inut

e in

terv

al

Absolute change in 15 minute volumes per site relative to base year 2001

0 1 2 6543 12 13 14 15 16 17 18 19 20 21 22 23 241110987

0 1 2 6543 12 13 14 15 16 17 18 19 20 21 22 231110987

99


Congestion estimation method–avoidable social costs

The cost estimates given in the above section (i.e. total time costs for congestion delay) are not actually very useful for assessing the impacts of congestion on society–or for valuing the avoidable social costs of congestion (i.e. those costs that could potentially be saved under appropriate policy or operational intervention).

As already mentioned, estimates of the ‘total cost of congestion delay’ are based on the value of the excess travel time (and other external or resource costs) incurred by the actual traffic over those that would have occurred had that traffic volume operated under completely free-flow conditions. Such conditions are of course an unrealisable hypothetical situation for actual road systems. So the congestion cost values given so far are really only useful as trend indicators, when comparing totals over time, rather than a direct measure of actual savings in social costs that may possibly be made through congestion-reduction measures.

A better set of values to use are the estimated deadweight losses (DWLs) associated with a particular congestion level–essentially giving a measure of the ‘cost of doing nothing about congestion’ or the ‘avoidable social costs of congestion’. That is, DWL values give an estimate of how much total social costs could be reduced if traffic volumes were reduced (either by appropriate pricing mechanisms or other demand management techniques) to the economically optimal level (i.e. to traffic volumes beyond which the full social costs of any further travel would outweigh the social benefits of that extra travel–given by the intersection of the travel demand curve and the generalised marginal travel cost curve).

For a brief overview of the theory behind the DWL valuation, and what portion of total congestion costs that the DWLs account for, refer to figure 2.37 below (reproduced BTCE 1995).

100


Figure 2.37 Basic economic theory of congestion costs

Source: BTCE (1995a).

The average cost curve in the diagram represents the unit cost of travel as perceived by individual road users and is essentially the basis for individual trip decision making. This curve is generally comprised of vehicle operating costs (such as maintenance and fuel expenses) and travel time costs (i.e. the average trip time, at that level of congestion, multiplied by a dollar value of time). Since the costs include dollar values for non-financial quantities (i.e. travel time) they are generalised costs.

This ‘Average Cost’ curve is given by averaging the total generalised travel cost for all road users (for the traffic level on the network link under consideration) across the number of those users (i.e. the product of the generalised unit cost at point A, equal to value V, by the quantity of travel for point A, equal to value F, gives the total cost incurred by all road users at that level of traffic). The marginal (social) cost curve is given by the derivative of this total cost (with respect to quantity of travel). That is, the ‘Marginal Cost’ curve gives the contribution to total travel costs for the marginal unit of travel, where the distance between the marginal and average cost curves represents the additional costs imposed on others (but not typically taken into account) by an extra motorist entering the traffic.

The total cost of congestion is equal to the size of the area bounded by the points VACT—or equivalently the area under the marginal cost

Generalised unit cost

Demand for travel

Marginal cost

Average cost

Quantity of Travel

W

V

U

T

O

P

A

DCS

R

Q

B

E FEconomic analysis of congestion

101


curve less the generalised costs at free flow (i.e. the area TPC). A better measure of congestion impacts is the ‘external’ costs of congestion—i.e. the costs that road users impose on others, through not having to personally meet the total costs caused by their travel decisions—given by the difference between the area under the marginal cost curve and the average cost curve, equal to area TPA. Though somewhat more sound than total delay costs (as an indicator of congestion’s actual social impact), the external costs of delay still tend to overstate the actual problems associated with congestion since they are still relative to the unrealisable situation of free flow travel speeds.

Theoretically, a better measure of congestion’s social costs is given by the deadweight loss of the current congestion level. The problem with a currently congested situation is that it includes a quantity of travel (in the range between the points E to F along the x-axis of Figure 2.37) for which the total travel costs exceed the social benefits of undertaking that travel. The net loss on this travel (the DWL of current congestion) is given by the area between the marginal cost curve and the demand curve (within the external cost area)–i.e. the area PAQ. Avoiding this loss (e.g. through the use of congestion charging or other transport demand management) thus provides a net social benefit of this amount (i.e. PAQ, evaluated in dollars), giving an estimate of the ‘cost of doing nothing about congestion’ or alternatively the ‘net benefit from optimal congestion reductions’. DWL estimates (termed here the ‘avoidable’ costs of congestion) have the advantage of not being the costs relative to the unfeasible situation of total free-flow, but rather are costs relative to the economically most efficient level of traffic for the road network.

Another (essentially equivalent) way viewing of the definition of such social costs (that is independent of any considerations of required pricing or traffic management mechanisms) consists of considering two of the main cost components that change as the traffic gets heavier. Referring again to Figure 2.37, consider the case of the traffic situation at point R (with quantity of travel E and average travel cost U) increasing to the level at point A (with traffic intensity F and average travel cost V). The two main changes to net social welfare involve:

• an increase in consumer surplus for the extra travellers (whose overall utility improves) by an amount given by the area BAQ; and

102


• an increase in total travel costs for all existing users (due to the higher congestion at point A), given by the area VBRU.

The net increase in costs from the increased traffic congestion is therefore equal to area VBRU less area BAQ, which given the geometry of the marginal cost curve, is equal to area PAQ. Thus if point R is taken as a suitable benchmark for congestion comparisons (since as it is defined here by the intersection of the travel demand curve and the marginal cost curve, the benefits of any extra travel beyond this point are outweighed by the extra costs to society), then area PAQ becomes the most economically appropriate value to take for the social costs of that congestion.

It should also be noted that even though these cost levels are theoretically avoidable, they do not directly relate to any net savings that may be possible under any particular congestion abatement policies. A DWL valuation gives an (order-of-magnitude) estimate of the worth society places on the disadvantages of current congestion-related delays and transport inefficiencies, relative to travel under less-dense traffic conditions (i.e. at economically optimal traffic levels). It does not allow for any of the wide range of costs that would be associated with actually implementing specific traffic management measures. The introduction of any measure aimed at congestion reduction will typically incur both set-up and ongoing operating costs, which will have to be considered separately from any benefits arising from changes in overall consumer utility (estimated here by the DWL reductions).

Also, remember that the derived congestion costs (dollar amounts) are not directly attributable to some equivalent proportion of GDP, since a major share of the congestion cost values refer to elements that play no part in the evaluation of GDP, such as private travel costs.

For the current analyses, average cost curve functions (and their associated marginal cost curves) have been derived for the various components of the cost calculations (largely based on the Bureau TRANSTEP modelling work, suitably calibrated for more recent aggregate network performance data). The calculation of the various cost elements (especially those given by the different areas just discussed with regards to figure 2.37) then proceeds using constant elasticity demand functions (derived from results given in BTCE Report 92) and typical aggregate relationships between the cost elements (again discussed in Report 92).

103


The resulting functional forms (in the current BTRE methodology) typically have the general shapes displayed in the following illustration (figure 2.38, which displays two different examples of possible demand curves). It should be noted that the current aggregate methodology relies on the assumption of constant demand elasticities—where the characteristic elasticity (with respect to total generalised travel cost) of the demand curves is taken to be -1.2, based on analyses described in Chapter 5 of Bureau Report 92 (BTCE 1996b). Other parts of the demand analyses use elasticities with respect to just the fuel price component of total travel costs, and these elasticities range between -0.1 to -0.3, depending on the type of vehicle use and the term over which the demand response is calculated. The estimated results would differ somewhat if the chosen demand elasticity was changed in value–or if the elasticity actually varies considerably between different periods of the day and different groups of travellers (i.e. the elasticity is highly non-constant).

Figure 2.38 Modelled average and marginal cost curves

For each road type in the model, the chosen speed-flow relationships (using the formula given on page 81 of the report) allow the calculation of average cost curves, which are then differentiated to obtain marginal cost curves. The derived cost functions are then integrated, to derive the various areas under the curves (corresponding to the cost elements discussed with regards to figure 2.37). For example, consider the following graph (figure 2.39), which plots the result of

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

High demand

Low demand

Average cost

Marginal cost

Trav

el ti

me

(min

s/km

)

Typical functional forms for travel costs

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Exte

rnal

pro

port

ion

of to

tal t

ime

dela

y co

sts

Typical functional form for external cost component

Volume-capacity ratio of road link

0.35 0.4 0.45 0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1.00 1.05

Volume-capacity ratio

104


calculating the areas TPA and TPC (again referring to the elements of figure 2.37) for the functional form of a typical divided arterial (given by using the parameter values of a = 16.7, b = 0.004 and T = 1 min/km in the formula given on page 81). The values in figure 2.39 are equal to the division of area TPC by area TPA, which provides an estimate of the proportion of the total costs that are external costs, for this example road link.

Figure 2.39 Proportion of external costs for a typical divided arterial

The next figure (2.40) then shows the typical proportion of total time costs that are external costs for the estimated network as a whole, over the hours of a typical week-day—i.e. after summing components from the major road types in the model, each divided according to their relevant external-proportion curves, such as given in Figure 2.39—again based on the results of the Bureau TRANSTEP modelling work and integration of the derived aggregate functional forms (e.g. those shown in figure 2.38).

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

High demand

Low demand

Average cost

Marginal cost

Trav

el ti

me

(min

s/km

)

Typical functional forms for travel costs

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Exte

rnal

pro

port

ion

of to

tal t

ime

dela

y co

sts

Typical functional form for external cost component

Volume-capacity ratio of road link

0.35 0.4 0.45 0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1.00 1.05

Volume-capacity ratio

105


Figure 2.40 Proportion of external costs

The demand curves (with the assumed constant elasticity) are then calibrated to pass through their relevant point A (again referring to the diagram of figure 2.37), so that their intersection with the derived marginal cost curve fixes the placement of point Q (and thus the size of area PAQ). Proceeding with a similar (numerical) integration to that for figure 2.39 yields a similar relationship for the DWL component (i.e. for what proportion area PAQ is of area TPA, depending on the volume-capacity ratio of the road link). Using this proportional relationship on the external cost values previously calculated (e.g. Figure 2.40) for each road type, and summing across all the modelled components yields, an aggregate estimate of the DWL fraction, and thus the following proportional makeup of total costs (figure 2.41), where the Figure shows the relative amounts typically accounted for by the external costs and the deadweight losses for current traffic profiles.

0

2

4

6

8

10

12

14

Hour of day

External component of total time costs

% o

f tot

al d

aily

cos

ts

0

2

4

6

8

10

12

14

External cost level

DWL of congestion delay

Total cost level

Hour of day

Components of external costs and deadweight losses

% o

f tot

al d

aily

cos

ts

Estimated external cost of congestion delay

Total travel time cost

0 1 2 6543 12 13 14 15 16 17 18 19 20 21 22 23 241110987

0 1 2 6543 12 13 14 15 16 17 18 19 20 21 22 23 241110987

106


Figure 2.41 Proportion of external costs and deadweight losses

The resulting average proportions from this estimation process (using the current model specification) are for aggregate external costs to be approximately 70 per cent of total costs (typically ranging between about 60 to 75 per cent) and for the deadweight losses to be around 50 per cent of total costs (typically ranging between about 30–55 per cent). It should be noted that this part of the estimation process is quite approximate, and is reliant on several input assumptions (e.g. the demand curve elasticity), and could introduce an additional element of uncertainty into the final cost estimates, perhaps of the order of 10–20 per cent.

Once the DWL values have been calculated for each city network (for each year), estimates of the full (avoidable) social costs of congestion are derived by adding in the appropriate cost estimates for extra vehicle operating cost (VOC) and for extra air pollution (using the types of functions displayed in figures 2.21 and 2.22).

The resulting daily composition of these aggregate avoidable costs (which essentially comprise the possible net benefit from optimal congestion reductions) is shown in figure 2.42.

0

2

4

6

8

10

12

14

Hour of day


% o

f tot

al d

aily

cos

ts

0

2

4

6

8

10

12

14

External cost level

DWL of congestion delay

Total cost level

Hour of day

Components of external costs and deadweight losses

% o

f tot

al d

aily

cos

ts

Estimated external cost of congestion delay

Total travel time cost

0 1 2 6543 12 13 14 15 16 17 18 19 20 21 22 23 241110987

0 1 2 6543 12 13 14 15 16 17 18 19 20 21 22 23 241110987

107


Figure 2.42 Typical daily profile for avoidable costs of congestion

Avoidable social cost values

BTRE base case results give a total of about $9.39 billion for national social costs of congestion (i.e. potentially avoidable costs, calculated on a DWL of current congestion basis) over the Australian capitals for 2005. This total value is comprised of approximately $3.5 billion in private time costs (DWL of trip delay plus variability), $3.6 billion in business time costs (DWL of trip delay plus variability), $1.2 billion in extra vehicle operating costs, and $1.1 billion in extra air pollution costs. The national total is spread over the capital cities with Sydney the highest (at about $3.5 billion), followed by Melbourne (with about $3.0 billion), Brisbane ($1.2 billion), Perth ($0.9 billion), Adelaide ($0.6 billion), Canberra ($0.11 billion), Hobart ($50 million) and Darwin ($18 million).

Projections of congestion costs

Using the base case projections for metropolitan VKT, in the congestion models described in the previous sections, then yields forecast values for the congestion cost estimates—taken to the year 2020 in the current study.

0.0

2.0

4.0

6.0

8.0

10.0

12.0

14.0

Hour of day

1 53 13 15 17 19 21 231197

Estimated external costs of congestion delay


% o

f tot

al d

aily

cos

ts

Increased vehicle operating costs

Extra air pollution

0

2

4

6

8

10

12

14

% o

f tot

al d

aily

cos

ts

Hour of day

Social costs of congestion

Travel time costs (difference between current congestion and ecomonic optimum)

0

100

200

300

400

500

600

700

800

Inde

x of

soc

ial c

osts

Hour of day

Growth in total projected congestion costs

Year 2020 costs

Year 2000 costs

0 1 2 6543 12 13 14 15 16 17 18 19 20 21 22 23 241110987

0 1 2 6543 12 13 14 15 16 17 18 19 20 21 22 23 241110987

108


The base case demand projections (coupled with the speed-flow, daily profile and marginal cost functions in the BTRE models) have the value of national metropolitan costs rising to an estimated $20.4 billion by 2020 (on an avoidable cost of congestion basis). Of this total, private travel is forecast to incur time costs of approximately $7.4 billion (DWL of trip delay plus variability), and business vehicle use $9 billion (DWL of trip delay plus variability). Extra vehicle operating costs contribute a further $2.4 billion and extra air pollution damages a further $1.5 billion. The city specific levels rise to approximately $7.8 billion for Sydney, $6.1 billion for Melbourne, $3.0 billion for Brisbane, $1.1 billion for Adelaide, $2.1 billion for Perth, $0.07 billion for Hobart, $35 million for Darwin, and $0.2 billion for Canberra.

The manner in which the average daily profile (for the avoidable social costs of congestion) changes over the projection period is shown in the figure 2.43.

Figure 2.43 Projected daily cost profile

Note: The congestion-delay elements of these plotted costs relates to the DWL component of total time costs.

The following tables (2.11 to 2.13) give the base case estimates for (avoidable) social costs, average network traffic performance, and average unit social costs due to urban traffic congestion (in the 8 Australian capital cities). The national social costs of metropolitan congestion for Australia (on a DWL basis) are also illustrated in the two charts following the time-series tables (figures 2.44 and 2.45).

0.0

2.0

4.0

6.0

8.0

10.0

12.0

14.0

Hour of day

1 53 13 15 17 19 21 231197

Estimated external costs of congestion delay


% o

f tot

al d

aily

cos

ts

Increased vehicle operating costs

Extra air pollution

0

2

4

6

8

10

12

14

% o

f tot

al d

aily

cos

ts

Hour of day


Travel time costs (difference between current congestion and ecomonic optimum)

0

100

200

300

400

500

600

700

800

Inde

x of

soc

ial c

osts

Hour of day

Growth in total projected congestion costs

Year 2020 costs

Year 2000 costs

0 1 2 6543 12 13 14 15 16 17 18 19 20 21 22 23 241110987

0 1 2 6543 12 13 14 15 16 17 18 19 20 21 22 23 241110987

109


Table 2.11 Base case projected estimates of social costs of congestion for Australian metropolitan areas, 1990–2020

(billion dollars)


1990 2.045 1.797 0.545 0.350 0.433 0.032 0.009 0.061 5.2721991 2.042 1.773 0.530 0.353 0.420 0.030 0.009 0.060 5.2181992 2.074 1.808 0.556 0.360 0.439 0.032 0.010 0.065 5.3421993 2.149 1.881 0.582 0.372 0.461 0.033 0.010 0.068 5.5561994 2.222 1.949 0.609 0.383 0.482 0.034 0.010 0.070 5.7601995 2.381 2.056 0.682 0.411 0.539 0.038 0.011 0.077 6.1951996 2.489 2.103 0.731 0.423 0.578 0.040 0.012 0.082 6.4601997 2.568 2.050 0.680 0.430 0.592 0.041 0.012 0.084 6.4571998 2.599 2.214 0.750 0.463 0.614 0.040 0.013 0.085 6.7771999 2.818 2.170 0.852 0.467 0.674 0.043 0.014 0.091 7.1292000 2.860 2.229 0.944 0.494 0.694 0.045 0.014 0.093 7.3742001 2.764 2.218 0.900 0.484 0.691 0.044 0.014 0.092 7.2072002 2.961 2.495 0.907 0.513 0.726 0.046 0.016 0.098 7.7622003 3.009 2.533 1.010 0.526 0.753 0.046 0.016 0.101 7.9942004 3.405 2.919 1.135 0.580 0.839 0.050 0.018 0.112 9.0572005 3.531 3.019 1.190 0.596 0.875 0.050 0.018 0.114 9.394

2006 3.692 3.125 1.256 0.616 0.919 0.051 0.019 0.117 9.7962007 3.967 3.340 1.361 0.652 0.991 0.053 0.020 0.124 10.5072008 4.259 3.571 1.472 0.690 1.065 0.055 0.021 0.131 11.2652009 4.577 3.826 1.593 0.731 1.145 0.057 0.023 0.138 12.0902010 4.871 4.054 1.709 0.767 1.223 0.059 0.024 0.145 12.8522011 5.133 4.253 1.817 0.799 1.294 0.060 0.025 0.151 13.5312012 5.392 4.447 1.926 0.829 1.365 0.061 0.026 0.156 14.2022013 5.649 4.637 2.040 0.858 1.440 0.062 0.027 0.161 14.8742014 5.915 4.832 2.160 0.888 1.517 0.063 0.028 0.167 15.5712015 6.189 5.032 2.286 0.919 1.598 0.064 0.029 0.172 16.2892016 6.463 5.227 2.414 0.949 1.680 0.065 0.030 0.177 17.0052017 6.762 5.442 2.557 0.982 1.772 0.066 0.031 0.183 17.7942018 7.068 5.653 2.702 1.014 1.864 0.067 0.032 0.189 18.5902019 7.401 5.880 2.859 1.048 1.963 0.068 0.034 0.195 19.4472020 7.755 6.123 3.027 1.084 2.068 0.069 0.035 0.201 20.362

Note: Time costs are based on deadweight losses for current congestion. That is, social costs refer here to the estimated aggregate costs of delay, trip variability, vehicle operating expenses and motor vehicle emissions—associated with traffic congestion—being above the economic optimum level for the relevant network. ‘Total’ column values refer to weighted averages over all metropolitan travel.


110


Table 2.12 Base case projections for average network delay due to congestion for Australian metropolitan areas, 1990–2020

(minutes per km)


1990 0.285 0.281 0.185 0.216 0.177 0.115 0.061 0.107 0.2431991 0.288 0.279 0.180 0.219 0.174 0.110 0.062 0.104 0.2431992 0.290 0.282 0.185 0.222 0.179 0.114 0.065 0.109 0.2461993 0.293 0.287 0.190 0.225 0.184 0.116 0.066 0.112 0.2501994 0.295 0.291 0.195 0.229 0.189 0.118 0.067 0.114 0.2531995 0.298 0.291 0.209 0.235 0.202 0.125 0.071 0.120 0.2581996 0.301 0.288 0.217 0.235 0.210 0.128 0.073 0.124 0.2601997 0.304 0.274 0.193 0.234 0.212 0.129 0.074 0.125 0.2551998 0.306 0.291 0.213 0.247 0.218 0.126 0.076 0.125 0.2651999 0.323 0.273 0.234 0.246 0.235 0.132 0.078 0.130 0.2702000 0.318 0.276 0.264 0.257 0.236 0.136 0.081 0.130 0.2742001 0.306 0.276 0.245 0.249 0.232 0.132 0.080 0.128 0.2662002 0.319 0.298 0.234 0.256 0.235 0.136 0.084 0.134 0.2772003 0.317 0.298 0.259 0.262 0.239 0.135 0.085 0.135 0.2802004 0.345 0.332 0.280 0.280 0.256 0.141 0.090 0.143 0.3062005 0.350 0.335 0.286 0.283 0.261 0.141 0.091 0.143 0.310

2006 0.358 0.342 0.294 0.288 0.267 0.142 0.093 0.145 0.3172007 0.374 0.356 0.308 0.299 0.279 0.145 0.096 0.150 0.3312008 0.390 0.370 0.323 0.309 0.291 0.148 0.099 0.154 0.3452009 0.407 0.386 0.338 0.321 0.304 0.151 0.101 0.159 0.3592010 0.421 0.399 0.352 0.330 0.315 0.153 0.104 0.163 0.3722011 0.433 0.409 0.364 0.338 0.325 0.155 0.106 0.166 0.3822012 0.444 0.419 0.375 0.344 0.333 0.156 0.108 0.169 0.3922013 0.454 0.428 0.385 0.351 0.342 0.157 0.110 0.171 0.4012014 0.464 0.436 0.396 0.357 0.350 0.158 0.112 0.174 0.4102015 0.475 0.445 0.407 0.363 0.359 0.158 0.113 0.176 0.4192016 0.484 0.453 0.418 0.369 0.367 0.159 0.115 0.179 0.4272017 0.495 0.462 0.429 0.375 0.376 0.160 0.117 0.181 0.4372018 0.505 0.470 0.440 0.381 0.384 0.161 0.118 0.184 0.4452019 0.516 0.479 0.452 0.388 0.393 0.161 0.120 0.186 0.4552020 0.527 0.488 0.464 0.393 0.402 0.162 0.122 0.188 0.464

Note: Relative to estimated free speeds.

Sources: Austroads (2006a), BTCE (1996a, b), BTRE estimates.

111


Table 2.13 Base case projections for average unit costs of congestion for Australian metropolitan areas, 1990–2020

(cents per PCU-km)


1990 6.4 6.2 4.3 4.1 3.9 2.1 1.3 2.2 5.41991 6.4 6.1 4.2 4.1 3.8 2.0 1.3 2.1 5.31992 6.4 6.2 4.3 4.2 3.9 2.1 1.3 2.2 5.41993 6.5 6.3 4.4 4.2 3.9 2.1 1.3 2.3 5.51994 6.6 6.4 4.5 4.3 4.0 2.2 1.4 2.3 5.51995 6.7 6.4 4.8 4.4 4.3 2.3 1.4 2.4 5.71996 6.8 6.4 5.0 4.4 4.4 2.3 1.5 2.5 5.71997 7.0 6.1 4.6 4.4 4.5 2.4 1.5 2.5 5.71998 6.9 6.5 4.9 4.6 4.6 2.3 1.5 2.5 5.81999 7.4 6.1 5.4 4.5 4.9 2.4 1.5 2.6 6.02000 7.3 6.2 5.9 4.7 4.9 2.5 1.6 2.6 6.02001 7.0 6.2 5.6 4.6 4.9 2.4 1.6 2.6 5.92002 7.2 6.6 5.4 4.7 4.9 2.5 1.6 2.7 6.12003 7.2 6.7 5.9 4.8 5.0 2.5 1.7 2.7 6.22004 7.8 7.3 6.3 5.1 5.3 2.6 1.7 2.9 6.72005 8.0 7.5 6.5 5.2 5.4 2.6 1.8 2.9 6.8

2006 8.2 7.6 6.7 5.2 5.6 2.6 1.8 3.0 7.02007 8.5 7.9 7.0 5.4 5.8 2.7 1.8 3.0 7.32008 8.9 8.2 7.3 5.6 6.0 2.7 1.9 3.1 7.62009 9.2 8.6 7.7 5.8 6.3 2.8 1.9 3.2 7.92010 9.6 8.9 8.0 6.0 6.5 2.8 2.0 3.3 8.22011 9.9 9.1 8.3 6.1 6.7 2.9 2.0 3.4 8.42012 10.2 9.4 8.6 6.2 6.9 2.9 2.1 3.5 8.72013 10.5 9.6 8.9 6.4 7.2 2.9 2.1 3.5 8.92014 10.8 9.9 9.2 6.5 7.4 3.0 2.1 3.6 9.22015 11.1 10.1 9.5 6.7 7.6 3.0 2.2 3.7 9.42016 11.3 10.3 9.8 6.8 7.8 3.0 2.2 3.7 9.72017 11.7 10.6 10.2 6.9 8.1 3.0 2.3 3.8 9.92018 12.0 10.8 10.5 7.1 8.4 3.1 2.3 3.8 10.22019 12.3 11.1 10.9 7.2 8.6 3.1 2.4 3.9 10.52020 12.7 11.4 11.3 7.4 8.9 3.1 2.4 4.0 10.8

Note: Time costs here are based on deadweight losses for current congestion. That is, these unit social costs refer to the estimated aggregate costs of delay, trip variability, vehicle operating expenses and motor vehicle emissions—associated with traffic congestion above the economic optimum level for the relevant networks—divided by the total (PCU-weighted) VKT on the network.


112


Figure 2.44 Projected avoidable costs of congestion by city

Note: Avoidable social costs are based on the deadweight losses associated with urban congestion levels (compared with the economically optimal traffic levels). Costs include congestion-related delays, trip variability, increased vehicle operating expenses and increased air pollution damages.

Sources: BTCE (1996b), BTRE estimates.

$bill

ion

Components of projected congestion costs

0

5

10

15

20

25

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

Private delay

Extra noxious pollutantsExtra vehicle operating costs

Trip variability

Business delay

Hobart

Perth

AdelaideBrisbane

Melbourne

Sydney

Canberra

Darwin

$bill

ion


0

5

10

15

20

25

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

113


Figure 2.45 Composition of projected avoidable costs of congestion

Note: Avoidable social costs are based on the deadweight losses associated with urban congestion levels (compared with the economically optimal traffic levels). Costs include congestion-related delays, trip variability, increased vehicle operating expenses and increased air pollution damages.

Sources: BTCE (1996b), BTRE estimates.

Uncertainty and sensitivity of estimation process

Estimating the ‘costs of congestion’ is typically a very approximate procedure. Due to a variety of methodological constraints and data limitations, any such cost estimates are far from definitive, forming more a rough (order of magnitude) picture, rather than an exact statement. In addition, the estimates will vary depending on the precise definitions used for the various ‘congestion’ effects—and to which conditions the congested traffic streams are considered relative. For example, current congested travel conditions could be compared to travel at average off-peak speeds, to hypothetical travel at posted speed limits or at free flow speeds, or to travel conditions at the economically optimal traffic level (which is determined by the intersection of the travel demand curve with the marginal generalised cost curve).

$bill

ion

Components of projected congestion costs

0

5

10

15

20

25

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

Private delay

Extra noxious pollutantsExtra vehicle operating costs

Trip variability

Business delay

Hobart

Perth

AdelaideBrisbane

Melbourne

Sydney

Canberra

Darwin

$bill

ion


0

5

10

15

20

25

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

114


Therefore, the estimate of how much total time is lost due to congestion—i.e. the exact value of delay that is derived—will depend directly on the ‘free’ speed values that the actual travel speeds of the congested traffic are compared with. Accurate details of free speeds (or average speeds for non-congested conditions) can be difficult to obtain–and different studies will often use quite different values for these comparison speeds (for evaluating delay).

As already pointed out, the current BTRE estimates use an aggregate methodology for estimating the costs of congestion (where the costs are based on the value of the excess travel time, and other external or resource costs, incurred by actual traffic levels over those that would have occurred had that traffic volume operated under more economically optimal conditions). Note that aggregate estimates of congestion delays (as used here) will generally underestimate the scale of the congestion problem (see BTCE Report 92, page 39, for some quantitative details on this issue). Traffic congestion typically occurs in a non-linear fashion—with only slight additions to vehicle numbers having the potential to radically lengthen travel times if the traffic stream is close to the road’s capacity—and aggregate methods will almost always have difficulty adequately capturing such responses.

It should be reiterated that accurate estimation of these effects typically requires the use of detailed traffic generation/assignment models, with all of a city’s main streets and traffic volumes encoded within the model. Given the extensive resources required to run (and continually update) such models for all of Australia’s capital cities, an aggregate method has been developed for this study, with the aim of deriving reasonable order-of-magnitude estimates for the national costs. However, such aggregate estimates should always be accompanied with the caveat that detailed network models (especially when used in combination with micro-simulation techniques) are far superior tools for calculating congestion effects and should be used whenever practicable. (See Austroads 2006a for a recent assessment of micro-simulation model use in Australia.)

Another source of possible cost underestimation relates to the question of how complete an enumeration of congestion externality effects is attempted. The costs presented in this report primarily relate to estimated externality values for excess travel time for road users (with some allowance for the inefficiency of vehicle engine

115


operation under stop-start conditions, leading to higher rates of fuel consumption and pollutant emissions). However, there will tend to be a series of other flow-on effects from urban congestion, ranging from some businesses having to re-locate or close (due to restrictions on their operations from congestion delays), to widespread psychological stress and irritation from coping with heavy traffic levels, to reducing the efficiency of public transit and the attractiveness of transit or non-motorised transport options. External costs of these wider effects will not be fully reflected in the current estimates and any attempt to include them would have been extremely approximate. Accurate valuations of such externalities are very difficult and there is, as yet, not a great deal in the literature covering suitable quantification methods. There is even considerable debate in the literature over the level that should be attached to the value of time lost, and the valuation techniques for ‘$/hr’ figures (generally based on average wage rates) are much more refined and accepted than those for such flow-on effects.

As well as these sources of possible model underestimation, there is also an issue of possible systematic overestimation. This has to do with the application of a standard, uniform value of time to practically all trips—whatever their origin-destination, purpose and possible level of urgency. The methodology does allow for some trips to be less time sensitive than others (about 10 per cent of trips in the base case assumptions), but it could actually be a much higher proportion. Some surveys suggest that many travellers (and in fact, possibly the bulk of off-peak travel and a substantial portion on non-business peak travel) do not place significant value on their delay time, as long as their chosen mode of transport is comfortable.

In summary, several of the largest sources of uncertainty (for delay cost estimates such as those of the present study) appear to concern:

• the level of free speeds used;

• the standard dollar value of time used;

• how many externalities, apart from delay costs, are included in the valuations; and

• how does the value of delay vary between different road users and their various trip purposes, times of travel or travel distances, especially regarding how much of urban travel is actually fairly time-sensitive.

116


As a guide to how widely the level of the derived congestion cost estimates are likely to vary, depending on model input values or methodological assumptions, a few sensitivity analyses are provided in this section of the report.

Firstly, a set of values significantly greater than the base case scenario (‘High Scenario 1’) is calculated:

• by including a rough allowance for the wider economic and social effects of congestion (based on the limited literature values available in this area), such as:

– costs to businesses (related to reductions in market accessibility, location choices, inventory practices, possible economies of scale, and labour/material supply options); and

– costs to non-car travellers (related to so-called ‘barrier effects’, where vehicle traffic and traffic congestion impose delays and discomfort on non-motorised modes (pedestrians and cyclists) and public transit (especially for trips involving a non-motorised component). Heavy traffic levels tend to reduce the viability of non-motorised travel, possibly leading to less than optimal modal choices, with associated external costs

• by removing the allowance for a proportion of total trips to be less time-sensitive than average (that is, the current base case setting, of approximately 10 per cent of trips assumed to have only half the standard value of time lost, is re-set to zero).

Then an even higher level scenario (‘High Scenario 2’) is obtained, by using the ‘High Scenario 1’ settings, and increasing assumed ‘free’ speeds for the networks by 10 per cent. These scenarios are shown in the figure 2.46.

Two related scenarios with lower cost levels than the base case have also been derived and are presented on the same graph (figure 2.46):

• ‘Low Scenario 1’ raises the proportion of total trips assumed to be less time-sensitive (to approximately half of all trips); and

• ‘Low Scenario 2’ is obtained from ‘Low Scenario 1’ by additionally decreasing assumed ‘free’ speeds for the networks by 10 per cent.

117


Figure 2.46 Sensitivity scenarios for major parameter variations

Note: ‘High Scenario 1’ adds a rough allowance for the wider economic and social costs of congestion (such as the costs to businesses related to reductions in possible economies of scale and the costs to non-motorised travel of barrier effects) and reduces the assumed proportion of time-insensitive trips. ‘High Scenario 2’ furthermore increases ‘free’ speeds assumed for the networks by 10 per cent. ‘Low Scenario 1’ raises the proportion of total trips assumed to be less time-sensitive. ‘Low Scenario 2’ furthermore decreases assumed ‘free’ speeds by 10 per cent.

Sources: NCHRP (2001), VTPI (2006), BTRE estimates.

$bill

ion

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

0

5

10

15

20

25

30

35High Scenario 1

High Scenario 2

Base case

Low Scenario 1

Low Scenario 2

Sensitivity ranges for congestion cost estimates: input assumptions

0

5

10

15

20

25

30High Scenario 3

Base case

Low Scenario 3

$bill

ion

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

Sensitivity ranges for congestion cost estimates: value of travel time

118


Figure 2.47 Sensitivity scenarios for variation in the value of time lost

Note: High sensitivity based on unit costs for value of time 50 per cent higher than base case. Low sensitivity based on unit costs for value of time 50 per cent lower than base case.


Probably the single greatest source of uncertainty in estimating social congestion costs relates to the assumed value of time—where there is considerable debate over exactly what ‘dollar per hour’ rate is actually most appropriate, for valuing the average traveller’s time lost to traffic delay, and how much that rate varies between different road users and different trip purposes. As an indicator of the possible uncertainty due to the value of time question, a further sensitivity test was performed by alternately raising the standard value of time—to be 50 per cent above the unit costs used in the base case (for ‘High Scenario 3’)—and then by lowering it to 50 per cent below the base value (‘Low Scenario 3’). These sensitivity tests are plotted in figure 2.47.

Yet another sensitivity test was conducted, to investigate the models’ responsiveness to input parameters concerning public transport patronage. A theoretical scenario was run on the model that assumed all non-motorised travel (walking and cycling) and all Urban Public Transit (UPT) passengers (rail and bus) were moved into private cars (for both past and projected urban travel).

$bill

ion

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

0

5

10

15

20

25

30

35High Scenario 1

High Scenario 2

Base case

Low Scenario 1

Low Scenario 2

Sensitivity ranges for congestion cost estimates: input assumptions

0

5

10

15

20

25

30High Scenario 3

Base case

Low Scenario 3

$bill

ion

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

Sensitivity ranges for congestion cost estimates: value of travel time

119


Surveys suggest that walking accounts for the order of 20 per cent of unlinked urban trips (that is, of all trip segments making up full journeys), since a walking portion is involved in much of urban travel (particularly at the start or end of a multi-modal journey). Yet trips mode-switchable to cars, such as those purely by walking or with walking segments involving a length greater than a kilometre, typically account for less than half of total walking distance. Since the average trip length for walking is quite short (typically in the range of 1 to 1.5 kilometres), walking trips substitutable by vehicle travel probably account for only around 1 per cent of total urban passenger-kilometres.

Figure 2.48 illustrates how the model’s cost estimates vary under the assumed response scenario for no (substitutable) trips by cycling, walking or urban public transport (i.e. all such trips are mode shifted to private road vehicles). The basic result of this rough analysis has the assumed mode shift (which moves approximately 12–13 per cent of aggregate urban pkm into private vehicles) increasing aggregate congestion costs by around a third (with about a 29 per cent increase in 2000, rising to about a 33 per cent divergence from the base case by 2020).

Figure 2.48 also plots the other side of this sensitivity test—that is, a model response scenario for a theoretical doubling of non-motorised and UPT travel. For this scenario, it was assumed that walking and cycling mode share could be doubled throughout the entire day—but that UPT travel would be subject to capacity constraints during the workday peak hours. The assumption was made that peak hour travel into the centre of the city, in the absence of extra UPT infrastructure provision, would typically only have around 20 per cent spare capacity. It was assumed that for all other times of the day, and for other trip types, that capacity constraints would not be a problem—and that the appropriate routing and marketing problems could be overcome, allowing a doubling of transit pkm task to be applied. The basic result of this further rough analysis sees the assumed mode shift (which has approximately 12 per cent of aggregate urban pkm switching out of cars) decreasing aggregate congestion costs by approximately 25 per cent in 2000, and approximately 27 per cent by 2020.

120


Figure 2.48 Sensitivity scenarios for UPT share variation

Figure 2.49 Sensitivity scenarios for travel demand variation

Figure 2.49 illustrates the final series of sensitivity tests conducted–primarily concerning input assumptions to the underlying transport demand modelling. The above chart (figure 2.49) shows the results for:

0

5

10

15

20

25

30Response scenario: no cycle, walk or UPT travelBase case

Response secenario: double base-case cycle, walk and UPT travel

$bill

ion

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

Sensitivity ranges for congestion cost estimates

0

5

10

15

20

25

30

35High demand (with trip variability cost) scenario

High demand scenario

Base case

Low demand scenario

Low demand (with low proportion of time-sensitive trips) scenario

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

$bill

ion


0

5

10

15

20

25

30Response scenario: no cycle, walk or UPT travelBase case

Response secenario: double base-case cycle, walk and UPT travel

$bill

ion

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020


0

5

10

15

20

25

30

35High demand (with trip variability cost) scenario

High demand scenario

Base case

Low demand scenario

Low demand (with low proportion of time-sensitive trips) scenario

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

$bill

ion


121


• a ‘High demand scenario’, which is based on inputting the highest likely population and economic growth over the projection period, coupled with minimal levels of future traffic peak spreading.

• a ‘Low demand scenario’ (based on the lowest likely population and economic growth over the projection period, coupled with maximal levels of future traffic peak spreading). The different peak-spreading assumptions make in the order of ±10 per cent difference to the 2020 aggregate cost levels.

• a ‘High demand (with high trip variability cost) scenario’, which is again based on inputs to the demand models of the highest likely population and economic growth over the projection period, coupled with minimal levels of future traffic peak spreading; along with significantly higher trip variability costs. The trip variability increase in this scenario results from using an 85th percentile for travel time spread (i.e. approximately 1.44 standard deviations from the mean, as opposed to 1 SD for the base case), and an elasticity of 2 for the value of time lost due to trip variability versus the value of time lost due to mean delay levels (as opposed to an elasticity value close to 1 for the base case).

• a ‘Low demand (with low proportion of time-sensitive trips) scenario’—again based on the lowest likely population and economic growth over the projection period, coupled with maximal levels of future traffic peak spreading; with the addition of a change to the assumption about average trip time urgency. Specifically, the base case assumption that only around 10 per cent of urban trips are relatively time-insensitive (and incur a value of time lost at half the standard dollar rate) is strengthened to allow for approximately 50 per cent of private trips to not incur appreciable time delay costs.

Current data deficiencies and possible future directions

The impacts of urban road congestion (vehicle delays and higher than average rates of fuel consumption and pollutant emissions) have been steadily increasing over time (due to the growing demand for vehicle

122


travel in our cities), with BTRE (base case) aggregate projections having the social costs of congestion possibly more than doubling over the 15 years between 2005 and 2020 (see table 2.11). The unit costs of congestion (that is, total avoidable congestion costs for metropolitan Australia divided by total VKT) are also forecast to rise, by around 59 per cent over this period (see table 2.13), as average delays become longer and congestion more widespread, and as freight and service vehicle use continues to grow strongly.

In summary, allowing for the sensitivity ranges discussed in the previous section, the costs imposed on Australian society by urban traffic congestion are likely to fall in the range of 5 to 15 billion dollars for current levels—in terms of theoretically avoidable costs (i.e. if, disregarding any possible implementation or running costs, appropriate traffic management or pricing schemes were to be introduced, so as to reduce traffic conditions to the economically optimal levels)—with a median value of around $10 billion. This is likely to rise, under base case demand growth assumptions, to a level of between $10 and $30 billion by 2020, with a median projected value for the potentially avoidable social costs of congestion of around $20 billion.

While preparing this study, the Bureau encountered a number of deficiencies in the readily available information on traffic volumes, and on consequent congestion conditions, that limited the costing analyses to various extents. Though various jurisdictions’ on-going Traffic Systems Performance Monitoring processes tend to collect a reasonable amount of useful congestion-related data (such as details on the distribution of traffic flows on the major road-links), there still appears to be a number of remaining data gaps—that could have a bearing on traffic and congestion management issues.

The BTRE’s assessment of the main data/information shortcomings, related to congestion assessment, is that they include:

• The relative scarcity of data details on freight movements and service vehicle usage—for both urban freight and services distribution and for freight volumes passing through urban areas. Though there are currently a few on-going studies looking into the issue of urban freight vehicle use, there are not yet much data available, e.g. on average truck operating characteristics, average loads and commodities carried, origin-destination patterns, or trip

123


scheduling. The lack of such information severely limits the suitable quantification of the impacts of congestion on freight costs.

• Limited data on the level and distribution of variability in average trip times. Though there are some data available on average network-wide variability magnitudes (mostly from limited floating-car measurements), there is little detailed information. That is, only scattered data seem to exist on how travel time variation is dependant on trip time of day and trip origin-destination; and its dependence on traffic volumes (especially as particular road-links approach full capacity). Trip time variability is particularly important to many businesses (especially those that need dependable service delivery times); and some travel surveys imply that many road users actually value travel time reliability considerably more highly than they value time lost to average delay levels.

• Typically only scattered data are available on the composition of the traffic mix and how it varies over the hours of the day. It would be useful if datasets were available that gave traffic volumes for significant portions of the networks in terms of vehicle type by time of day. Any further information that can be collected regarding the distribution of urban travel (e.g. by trip purposes or by journey origin-destinations) will also be useful for congestion modelling exercises.

As greater quantities of data continue to be collected in real-time, there would undoubtedly be social benefits in the greater provision of real-time traffic information systems, e.g internet sites showing current traffic performance statistics to aid travellers’ trip planning, allowing trips to be rescheduled or re-routed (if notified of a particular incident or irregular delay on their usual route).

Enhanced data collection processes, along with greater data accessibility, will not only aid transport decision making and the direct evaluation of congestion management practices, but will also allow more robust modelling of congestion occurrence and its social impacts–which will, in turn, further assist the policy evaluation process. There are a variety of modelling approaches currently used for estimating congestion levels and their associated costs—some studies (including this current analysis) use aggregate methods, which are computationally straightforward but very approximate; others

124


are based on detailed network models, which allow more precise traffic specifications but are much more data-intensive and difficult to adequately calibrate. The relative simplicity of aggregate methods means that, generally speaking, they are readily calibrated, and the current BTRE models have been calibrated against aggregate network performance data collected by the various State road authorities (including the annual statistics reported to the Austroads National Performance Indicators).

This allows current aggregate congestion cost levels (for the BTRE estimates) to be suitably benchmarked against actual on-road conditions, but it has to be acknowledged that such aggregate methods are rather blunt instruments for projecting congestion occurrence. Of course, aggregate methods are also not capable of describing the specific location or spatial distribution of the congestion over a particular city’s network. Detailed traffic congestion forecasts are much more accurately performed using suitably calibrated network models, and the Bureau looks forward to each jurisdiction’s efforts in developing network modelling tools that are continually more comprehensive, consistent, and informative.

For example, consider the Melbourne Integrated Transport Model (MITM), a detailed model of the Melbourne transport network used by the Victorian Department of Infrastructure (DOI). MITM has the advantage of allowing precise specifications of road link characteristics and the spatial assignment of the modelled traffic across those road links, and has been used to derive aggregate congestion cost estimates (e.g. see table 3.1 of VCEC 2006). However, MITM does not currently include any allowance for business traffic costs (VCEC 2006, pg. 63). The limitation of the modelled traffic to only passenger vehicles has a considerable bearing on comparability with some other studies, given that the current BTRE estimates typically have business costs as a greater proportion of the total congestion costs than private travel costs. The DOI continue to develop the MITM, and work is underway to redress this current shortcoming, with studies being undertaken aimed at adding an accurate representation of Melbourne’s freight and service vehicle flows to the model framework. Most Australian researchers working in the field of traffic and congestion modelling will undoubtedly be interested in the results of this work.

125


Though various Australian studies have tended to derive significantly varying congestion cost estimates, much of the variation can usually be explained by definitional differences (e.g. exactly which congestion effects are included in the costings) or by different parameter inputs (such as the value of time or the assumed level of free-flow speeds). As mentioned previously, the most significant factor in correctly determining the level of congestion costs tends to be accurately assigning a dollar value of time—where there is a high level of uncertainty in how adequately standard values of time capture the worth that travellers actually attach to delays, and there are little data available on how the time values vary over different trip types and trip timings.

In closing, observe that irrespective of the questions over exact dollar valuations raised by the sensitivity tests or modelling result caveats, the principal finding of this study remains: that, in the absence of improved congestion management, rising traffic volumes in the Australian capitals are likely to lead to escalating congestion impacts, such that the net social costs of congestion over the next 15 years (under a business-as-usual scenario) are likely to at least double.

Given the scale of the congestion problem, it will be a challenge for future transport management measures to adequately address the increasing trends identified by this study. The most effective of any measures introduced to combat growing congestion levels will probably be those:

• that most successfully engage the community (and thus combat any reluctance to change existing travel behaviour), and

• that target particular city areas or particular times of travel where congestion is highest (e.g. freeway tolls or continuous electronic road pricing).

Ideally, the public would be clearly shown how the costs of any proposed congestion reduction policies are outweighed by the benefits (such as reduced trip delays and less air pollution), that any charges imposed would relate directly to the extent the road user contributes to overall congestion levels (such as through optimal road pricing), and that the revenue raised by any pricing mechanisms be distributed in a socially acceptable manner.

Appendix Aggregate inputs and supplementary results

Table A.1 State and territory population projections

(thousand persons)

Year NSW Vic Qld SA WA Tas NT ACT Total

1990 5 832 4 378 2 905 1 431 1 615 462 164 283 17 0701991 5 896 4 415 2 967 1 444 1 637 466 166 289 17 2801992 5 955 4 445 3 039 1 452 1 658 469 168 295 17 4801993 6 005 4 464 3 123 1 459 1 679 471 171 299 17 6701994 6 056 4 478 3 194 1 461 1 705 472 173 301 17 8411995 6 120 4 507 3 270 1 464 1 735 472 178 304 18 0501996 6 242 4 585 3 365 1 479 1 779 476 184 309 18 4201997 6 280 4 612 3 409 1 479 1 803 472 188 307 18 5501998 6 328 4 650 3 453 1 482 1 831 470 190 307 18 7111999 6 399 4 703 3 505 1 490 1 857 469 193 310 18 9262000 6 472 4 763 3 560 1 496 1 887 468 195 312 19 1532001 6 555 4 827 3 624 1 506 1 919 468 199 315 19 4132002 6 626 4 878 3 686 1 513 1 949 468 202 318 19 6412003 6 687 4 917 3 797 1 527 1 952 477 198 323 19 8792004 6 731 4 973 3 882 1 534 1 982 482 200 324 20 1092005 6 845 5 019 3 892 1 538 2 047 469 213 328 20 3502006 6 909 5 057 3 957 1 543 2 077 469 217 331 20 5602007 6 973 5 095 4 022 1 549 2 108 468 220 333 20 7682008 7 039 5 135 4 090 1 555 2 140 468 224 336 20 9862009 7 103 5 173 4 156 1 560 2 171 468 228 339 21 1962010 7 167 5 211 4 223 1 566 2 202 467 231 341 21 4082011 7 229 5 246 4 288 1 570 2 233 467 235 344 21 6122012 7 287 5 280 4 353 1 574 2 263 466 238 346 21 8062013 7 343 5 310 4 415 1 577 2 292 464 242 348 21 9912014 7 398 5 341 4 478 1 580 2 321 463 245 350 22 1782015 7 455 5 372 4 542 1 584 2 351 462 249 353 22 3672016 7 508 5 401 4 604 1 586 2 379 460 253 355 22 5462017 7 561 5 430 4 667 1 589 2 408 459 256 357 22 7262018 7 611 5 456 4 727 1 591 2 436 457 260 358 22 8972019 7 661 5 482 4 788 1 593 2 464 455 263 360 23 0682020 7 712 5 509 4 850 1 595 2 493 453 267 362 23 241

Sources: BTRE estimates based on ABS (mid-range series B) long-term projections.

127

Table A.2 Capital city population projections

(thousand persons)

Year NSW Vic Qld SA WA Tas NT ACT Total

1990 3 632 3 127 1 331 1 047 1 174 189 75 283 10 8571991 3 672 3 153 1 359 1 056 1 189 190 76 289 10 9861992 3 712 3 180 1 390 1 063 1 207 192 77 295 11 1151993 3 746 3 199 1 426 1 068 1 225 193 78 299 11 2331994 3 782 3 214 1 456 1 070 1 246 194 78 301 11 3411995 3 825 3 241 1 489 1 072 1 270 194 80 304 11 4751996 3 904 3 302 1 530 1 084 1 305 196 82 309 11 7141997 3 940 3 327 1 550 1 083 1 324 195 84 307 11 8111998 3 978 3 364 1 573 1 086 1 342 194 86 307 11 9291999 4 033 3 410 1 598 1 091 1 361 194 88 310 12 0852000 4 093 3 462 1 624 1 097 1 385 193 90 312 12 2562001 4 157 3 516 1 654 1 106 1 410 193 91 315 12 4442002 4 212 3 561 1 684 1 113 1 433 193 93 318 12 6072003 4 270 3 604 1 715 1 121 1 457 193 95 323 12 7782004 4 325 3 643 1 747 1 128 1 481 194 97 324 12 9382005 4 382 3 682 1 780 1 135 1 505 194 99 328 13 106

2006 4 433 3 717 1 811 1 140 1 528 194 101 331 13 2542007 4 483 3 751 1 842 1 145 1 551 194 103 333 13 4032008 4 536 3 787 1 873 1 151 1 574 194 105 336 13 5572009 4 587 3 821 1 904 1 156 1 597 194 107 339 13 7062010 4 639 3 855 1 936 1 161 1 620 194 109 341 13 8572011 4 689 3 888 1 967 1 165 1 643 194 111 344 14 0022012 4 738 3 919 1 997 1 169 1 665 194 114 346 14 1422013 4 784 3 949 2 027 1 173 1 687 194 116 348 14 2772014 4 831 3 978 2 057 1 177 1 709 194 118 350 14 4122015 4 878 4 008 2 087 1 180 1 730 194 120 353 14 5492016 4 924 4 035 2 116 1 183 1 752 193 122 355 14 6802017 4 970 4 064 2 146 1 187 1 773 193 124 357 14 8122018 5 014 4 090 2 175 1 189 1 794 192 126 358 14 9382019 5 058 4 116 2 204 1 192 1 814 192 128 360 15 0652020 5 103 4 143 2 233 1 195 1 835 192 130 362 15 193

Sources: BTRE estimates based on ABS (mid-range series B) long-term projections.

128


Table A.3 Base case GDP growth assumptions

(per cent change per annum)

Financial year Australian real GDP growth

2003 3.212004 4.092005 3.002006 3.252007 3.502008 3.502009 3.102010 3.002011 2.852012 2.702013 2.552014 2.402015 2.302016 2.252017 2.202018 2.152019 2.102020 2.00

Source: AGO (pers. comm., based on Treasury estimates).

Figure A.1 Long-term trends in national per capita passenger and freight movement relative to per capita income levels


2001

2003

1999

1997

1995

1993

1991

1989

1987

1985

1983

1981

1979

1977

1975

1973

1971

0

5

10

15

20

25

30

35

40City Freight—Predicted

City Freight

billi

on to

nne-

kilo

met

res

0

5

10

15

20

25

5 10 15 20 25 30 35 40 45

Freight task–tkm per person

Passenger task–pkm per person

Per c

apita

tran

spor

t tas

k(th

ousa

nd p

km o

r tkm

per

per

son)

Per capita income (GDP/population)—thousand dollars

129

Appendix

Table A.4 Base case projected estimates of social costs of congestion for business vehicle travel, 1990–2020

($billion)


1990 1.023 0.908 0.266 0.150 0.202 0.011 0.005 0.022 2.5861991 1.012 0.889 0.255 0.150 0.193 0.010 0.005 0.021 2.5351992 1.029 0.907 0.268 0.152 0.202 0.010 0.005 0.023 2.5961993 1.066 0.944 0.282 0.157 0.212 0.011 0.005 0.024 2.7001994 1.104 0.979 0.296 0.162 0.223 0.011 0.006 0.025 2.8061995 1.188 1.035 0.336 0.175 0.252 0.012 0.006 0.028 3.0331996 1.242 1.060 0.363 0.180 0.272 0.013 0.007 0.030 3.1671997 1.284 1.032 0.335 0.183 0.279 0.013 0.007 0.031 3.1641998 1.300 1.116 0.373 0.196 0.291 0.013 0.007 0.031 3.3271999 1.421 1.093 0.427 0.198 0.322 0.014 0.008 0.034 3.5172000 1.434 1.118 0.475 0.208 0.329 0.015 0.008 0.034 3.6212001 1.379 1.112 0.451 0.204 0.327 0.014 0.008 0.034 3.5282002 1.485 1.260 0.454 0.216 0.345 0.015 0.009 0.036 3.8192003 1.504 1.276 0.507 0.221 0.356 0.015 0.009 0.037 3.9262004 1.707 1.478 0.571 0.243 0.397 0.016 0.010 0.042 4.4632005 1.774 1.534 0.601 0.251 0.415 0.016 0.010 0.043 4.643

2006 1.865 1.593 0.637 0.260 0.438 0.016 0.011 0.044 4.8652007 2.018 1.713 0.696 0.277 0.476 0.017 0.011 0.047 5.2562008 2.179 1.842 0.757 0.295 0.515 0.018 0.012 0.050 5.6682009 2.353 1.983 0.824 0.314 0.557 0.019 0.013 0.053 6.1152010 2.518 2.113 0.890 0.332 0.598 0.019 0.013 0.056 6.5402011 2.667 2.227 0.951 0.348 0.637 0.020 0.014 0.059 6.9232012 2.816 2.340 1.014 0.363 0.676 0.020 0.015 0.061 7.3052013 2.964 2.452 1.079 0.378 0.717 0.021 0.015 0.064 7.6902014 3.120 2.567 1.149 0.394 0.760 0.021 0.016 0.066 8.0932015 3.280 2.686 1.222 0.410 0.804 0.022 0.017 0.069 8.5092016 3.442 2.803 1.297 0.426 0.851 0.022 0.017 0.071 8.9292017 3.621 2.933 1.381 0.444 0.903 0.022 0.018 0.074 9.3972018 3.804 3.061 1.467 0.461 0.955 0.023 0.019 0.077 9.8682019 4.004 3.200 1.561 0.480 1.012 0.023 0.019 0.080 10.3802020 4.239 3.366 1.669 0.503 1.078 0.024 0.020 0.084 10.982

Note: Time costs are based on deadweight losses for current congestion. That is, social costs refer here to the estimated aggregate costs of delay, trip variability, vehicle operating expenses and motor vehicle emissions—associated with traffic congestion—being above the economic optimum level for the relevant network.


130


Table A.5 Base case projected estimates of social costs of congestion for private vehicle travel, 1990–2020

($billion)


1990 1.023 0.889 0.279 0.199 0.231 0.021 0.004 0.039 2.6861991 1.030 0.884 0.275 0.204 0.227 0.020 0.004 0.039 2.6821992 1.045 0.901 0.288 0.208 0.237 0.021 0.004 0.042 2.7461993 1.083 0.938 0.300 0.215 0.248 0.022 0.005 0.044 2.8551994 1.118 0.970 0.313 0.221 0.259 0.023 0.005 0.045 2.9541995 1.193 1.020 0.346 0.236 0.287 0.025 0.005 0.049 3.1621996 1.247 1.043 0.368 0.243 0.306 0.027 0.005 0.052 3.2931997 1.284 1.017 0.345 0.247 0.313 0.027 0.006 0.054 3.2921998 1.299 1.097 0.377 0.267 0.323 0.027 0.006 0.054 3.4501999 1.397 1.077 0.425 0.269 0.352 0.029 0.006 0.057 3.6132000 1.427 1.111 0.469 0.286 0.365 0.031 0.006 0.059 3.7542001 1.385 1.106 0.449 0.281 0.364 0.030 0.006 0.058 3.6792002 1.476 1.235 0.453 0.296 0.381 0.031 0.007 0.062 3.9432003 1.505 1.257 0.503 0.305 0.397 0.031 0.007 0.064 4.0682004 1.699 1.441 0.564 0.337 0.442 0.034 0.008 0.070 4.5942005 1.757 1.485 0.590 0.346 0.460 0.034 0.008 0.072 4.752

2006 1.828 1.532 0.618 0.356 0.481 0.034 0.008 0.073 4.9312007 1.950 1.626 0.665 0.374 0.514 0.036 0.009 0.077 5.2522008 2.081 1.729 0.715 0.395 0.550 0.037 0.009 0.081 5.5962009 2.224 1.842 0.769 0.417 0.589 0.038 0.010 0.085 5.9752010 2.353 1.941 0.820 0.435 0.625 0.039 0.010 0.089 6.3122011 2.466 2.025 0.866 0.451 0.657 0.040 0.011 0.092 6.6092012 2.576 2.106 0.912 0.466 0.690 0.041 0.011 0.095 6.8972013 2.684 2.185 0.961 0.480 0.723 0.041 0.012 0.098 7.1842014 2.795 2.265 1.011 0.495 0.758 0.042 0.012 0.100 7.4782015 2.909 2.346 1.064 0.509 0.794 0.043 0.012 0.103 7.7802016 3.021 2.424 1.117 0.523 0.829 0.043 0.013 0.106 8.0762017 3.141 2.509 1.175 0.538 0.869 0.044 0.013 0.109 8.3982018 3.264 2.591 1.235 0.553 0.909 0.044 0.014 0.112 8.7212019 3.396 2.679 1.298 0.568 0.951 0.045 0.014 0.115 9.0672020 3.516 2.757 1.358 0.581 0.990 0.045 0.015 0.117 9.380

Note: Time costs are based on deadweight losses for current congestion. That is, social costs refer here to the estimated aggregate costs of delay, trip variability, vehicle operating expenses and motor vehicle emissions—associated with traffic congestion—being above the economic optimum level for the relevant network.


131

Appendix

Figure A.2 Summary flowchart–BTRE base case projections for traffic growth

Data Input:Projected Population trends—national and by state,Urban and non-urban; Demographic trends;Economic trends and projected GDP growth;Fuel price trends;Other prices (e.g. vehicle prices, public transit fares, airfares, freight rates)…

BTRE demand models–Econometric components

Aggregate transport task projections—first iteration (unconstrained demand)

BTRE demand models–Modal split and competitiveness components

Modal transport demand projections—first iteration (unconstrained demand)

Infrastructure availability

Modelled saturation parameters and Feedback effects.

For example:• Increasing demand → increasing traffic congestion →

increasing generalised travel costs →some modal change and some limitation of original demand growth.

• Increasing fuel prices → increasing generalised travel costs

→ decreases in VKT and changes in vehicle choices→ eventual improved average fleet fuel efficiency→ some

reductions in generalised travel cost → some VKT increases.

Current and planned vehicle design specifications (Euro standards, ADRs, etc);Current and planned fuel quality standards;Emission testing data (including speed dependencies)...

BTRE fleet emission models

BTRE demand models— Structural components

After equilibrium (using generalised costs)

Modal transport demand projections

Vehicle fleet characteristics,Average scrappage rates,New vehicle fuel efficiency trends...

Iterate model inputs and outputs until equilibrium (for the feedback effects) is reached...

BTRE vehicle fleet models

Base case Motor Vehicle projections:For national and state-by-state fleets (including metropolitan and non-metropolitan estimates);VKT and fuel use by vehicle type, fuel type and age of vehicle.

Base case projections of VKT and vehicle fuel use, along with consequent emissions of greenhouse gases and noxious pollutants (all major species) by vehicle type, for each capital city.

132


Figure A.3 Summary flowchart–BTRE method for congestion cost estimation

Proportion of total time costs that are external and DWL

Derived average (generalised) cost curves and marginal cost curves

Inputs from Transport Demand Models:Projected annual VKT to 2020, by vehicle type and city;Speed dependency curves for emission rates (by gas species), average fuel consumption and vehicle operating costs.

Hourly VKT and PCU-km, by vehicle type, for an average weekday, for each major road type, for each capital city

Hourly passenger-km, by vehicle type, for an average weekday, for each major road type, for each capital city

Average daily traffic profiles for each vehicle type (hourly percentage of total);Difference in traffic levels between average weekday and weekend/public holiday.

Splitting factors / daily traffic profiles for each major road type

Average vehicle occupancy rates, by hour of the day;Non-business vs business proportion of travel, by vehicle type.

Projected aggregate delay, by vehicle type and hour of day

Projected changes in average network travel time

Assumed Values of time, by vehicle type

By road type:Speed-flow curves;Free/uncongested speeds;Base volume-capacity ratios.

Trip variability-traffic flow curves

Capacity trend assumptions

Projected changes in average speeds for each major road type;V-C ratio trends;Proportion of total emissions and vehicle operating costs due to congestion.

Calibrate demand curves to pass through current average costs, and determine intersection points with marginal cost curves.

Total time costs: for private vehicle delay, business vehicle delay and trip variability

‘Avoidable’ delay costs: for private vehicles and business vehicles

Avoidable Social Costs of Congestion: for extra air pollution, vehicle operating costs and travel time, for each capital city.

Sum over road types

Sum over vehicle types

133

Appendix

Figure A.4 Forecast fuel price trend for BTRE base case demand projections

Note: Crude oil prices are given here in 2004 US dollars (per barrel), based on US Energy Information Administration (EIA) results for their latest ‘Reference’ long-term projection scenario–adjusting the quoted $US RAC values (for calendar year forecasts) to WTI (2004 real) financial year values, averaging the given trends for all crudes (including light, low sulphur crudes), and smoothing across the latest EIA short-term forecasts

Sources: BTRE estimates based on US Energy Information Administration forecasts (Energy Information Administration 2006a).

Financial year

Cru

de o

il pr

ice

($U

S/br

l)

0

10

20

30

40

50

60

70

1990

1992

1998

2010

1994

2002

2000

2004

2006

2008

2012

2014

2016

1996

2018

2020

134


References

ABS Australian Bureau of Statistics AGPS Australian Government Publishing ServiceACG Apelbaum Consulting GroupAGO Australian Greenhouse OfficeBTCE Bureau of Transport and Communications EconomicsBTE Bureau of Transport EconomicsBTRE Bureau of Transport and Regional EconomicsCOAG Council of Australian GovernmentsDISR Department of Industry, Science and ResourcesDITR Department of Industry, Tourism and ResourcesDOTARS Department of Transport and Regional ServicesDPIE Department of Primary Industries and EnergyFORS Federal Office of Road SafetyNGGIC National Greenhouse Gas Inventory CommitteeNCHRP National Cooperative Highway Research ProgramUSEPA United States Environmental Protection AgencyVCEC Victorian Competition and Efficiency CommissionVTPI Victoria Transport Policy Institute

ABS (1993), Motor Vehicle Registrations, Australia, Cat. no. 9304.0, ABS, Canberra.

ABS (1996), Survey of Motor Vehicle Use, Australia, 30 September 1995, Preliminary, Cat. no. 9208.0, ABS, Canberra.

ABS (2000a), Survey of Motor Vehicle Use, Australia, 12 months ended 31 July 1998, Cat. no. 9208.0, ABS, Canberra.

ABS (2000b), Motor Vehicle Census, Australia, Cat. no. 9309.0, ABS, Canberra.

ABS (2000c), Survey of Motor Vehicle Use, Australia, 12 months ended 31 July 1999, Cat. no. 9208.0, ABS, Canberra.

ABS (2001a), Survey of Motor Vehicle Use, Australia, 12 months ended 31 October 2000, Cat. no. 9208.0, ABS, Canberra.

135

ABS (2001b), Australian Demographic Statistics, Cat. no. 3101.0, ABS, Canberra.

ABS (2003), Survey of Motor Vehicle Use, Australia, 12 months ended 31 October 2001, Cat. no. 9208.0, ABS, Canberra.

ABS (2005), Australian Demographic Statistics, Cat. no. 3101.0, ABS, Canberra.

Akcelik, R. (1978), ‘A new look at Davidson’s travel time function’, Traffic Engineering and Control, vol.19.

Akcelik, R. (1991), ‘Travel time functions for transport planning purposes: Davidson’s function, its time-dependent form and an alternative time function’, Australian Road Research, vol.21, no.3.

Austroads (2004), Economic Evaluation of Road Investment Proposals - Unit Values for Road User Costs at June 2002, Report AP-RS41.

Austroads (2006), The use and application of microsimulation traffic models, Austroads Report No. AP-R286/06, prepared by James Luk and Johann Tay.

BTCE (1995a), Urban Congestion: Modelling Traffic Patterns, Delays and Optimal Tolls, Working Paper 15, BTCE, Canberra.

BTCE (1995b), Greenhouse Gas Emissions from Australian Transport — Long-term Projections, Report 88, BTCE, Canberra.

BTCE (1996a), Transport and Greenhouse: Costs and options for reducing emissions, BTCE Report 94.

BTCE (1996b), Traffic Congestion and Road user chargers in Australian capital cities, BTCE Report 92.

BTCE (1997), Roads 2020, Working Paper 35, BTCE, Canberra.

BTE (1999), Urban Transport – Looking Ahead, BTE Information Sheet 14.

BTE (2000), Urban Congestion - The Implications for Greenhouse Gas Emissions, Information Sheet 16.

BTRE (2002a), Greenhouse Gas Emissions from Transport – Australian Trends to 2020, BTRE Report 107.

BTRE (2002b), Freight Rates in Australia, Information Sheet 19, BTRE, Canberra.

136


BTRE (2003a), Urban Pollutant Emissions From Motor Vehicles: Australian Trends To 2020, Final Draft Report for Environment Australia, June 2003; study conducted by D. Cosgrove for the Department of Environment and Heritage, BTRE. <http://www.btre.gov.au/docs/joint_reports/urbanpollutants_draft.aspx>

BTRE (2003b), Aggregate Greenhouse Gas Emissions from Australian Transport: Base Case Projections (Bottom-Up Approach) to 2020, unpublished consultancy report to AGO, July 2003.

BTRE (2005), Health Impacts of Transport Emissions in Australia: Economic Costs, Working Paper 63, DOTARS, Canberra.

BTRE (2006a), Greenhouse Gas Emissions from Australian Transport: Base Case Projections to 2020, Report for AGO, August 2005, < http://www.btre.gov.au/docs/commissioned/BTRE_AGO_05.pdf>

BTRE (2006b), Freight Measurement & Modelling in Australia, BTRE Report 112, Canberra.

COAG (2006), February 2006 communiqué, <http://www.coag.gov.au/meetings/100206/index.htm#reform>

Energy Information Administration (2006), Annual Energy Outlook 2006, AOE2006 - Annual Energy Outlook 2005 with Projections to 2030, <www.eia.doe.gov/oiaf/aeo/index.html>

Kimber & Hollis(1979), Traffic Queues and Delays at Road Junctions, Report 909, Transport and Road Research Laboratory, UK.

NCHRP (2001), Economic Implications of Congestion, NCHRP Report 463, Transportation Research Board, Washington DC.

Treasury (2002), Australia’s long-term demographic and economic prospects, Part II, Budget Paper No. 5, Canberra.

VCEC (2006), Making the right choices: Options for managing transport congestion, Draft report, April 2006, State of Victoria, Melbourne.

VicRoads (2001), Traffic System Performance Monitoring 1999/2000, VicRoads, Melbourne.

VicRoads (2005), VicRoads Information Bulletin, Traffic Systems Performance Monitoring 2003/2004, Melbourne.

137

References

VicRoads (2006), VicRoads Information Bulletin, Traffic Systems Performance Monitoring 2004/2005, Melbourne.

VTPI (2006), http://www.vtpi.org/documents/transportation.php

Watkiss, P. 2002, Fuel Taxation Inquiry: The Air Pollution Costs of Transport in Australia.

138


Further readings

ABS (2002), Motor Vehicle Census, Australia, 31 March 2002, Cat. no. 9309.0, ABS, Canberra.


ABS (2004b), Freight Movements, Australian Summary, March 2001, Cat. no. 9220.0, ABS, Canberra.

ABS (2004c), Motor Vehicle Census, Australia, 31 March 2004, Cat. no. 9309.0, ABS, Canberra.


ABS (2005b), Australian National Accounts: National Income, Expenditure and Product, Cat. no. 5206.0, ABS, Canberra.

ABS (2005c), Sales of New Motor Vehicles, Australia, Cat. no. 9314.0.55.001, ABS AusSTATS.

ACG (1993), The Australian Transport Task and the Primary Energy Consumed—1990/91 Compendium, Vol. B (study commissioned by DPIE, BTCE, AAA, AIP Ltd., ESAA Ltd.), DPIE, Canberra.

ACG (1997), The Australian Transport Task, Energy Consumed and Greenhouse Gas Emissions—1994/95 Compendium, Vol. B (study commissioned by DPIE, BTCE, AAA, AIP Ltd., ESAA Ltd.), DPIE, Canberra.

ACG (2001), Australian Transport Facts 1998, Report prepared for DISR, Apelbaum Consulting Group, Eastern Press Pty Ltd, Melbourne.

ACG (2005), Australian Transport Facts 2003, Apelbaum Consulting Group Pty Ltd, Melbourne.

Austroads (2006), Austroads National Performance Indicators (www.austroads.com.au).

139

BTCE (1991), Short-term Forecasting of Transport and Communications Activity, Working Paper 2, BTCE, Canberra.

BTCE (1995a), Costs of Reducing Greenhouse Gas Emissions from Australian Road Freight Vehicles: An application of the BTCE TRUCKMOD model, Working Paper 22, BTCE, Canberra.

BTCE (1995b), Costs of Reducing Greenhouse Gas Emissions from Australian Cars: An application of the CARMOD model, Working Paper 24, BTCE, Canberra.

BTE (1998), Forecasting Light Vehicle Traffic, Working Paper 38, BTE, Canberra.

BTE (1999a), Analysis of the Impact of the Proposed Taxation Changes on Transport Fuel Use and the Alternative Fuel Market, Unpublished consultancy report for Environment Australia, prepared by D. Cosgrove, April 1999, BTE, Canberra.

BTE (1999b), Trends in Trucks and Traffic, Information Sheet 15, BTE, Canberra.

BTRE (2002), Fuel consumption by new passenger vehicles in Australia, Information Sheet 18, BTRE, Canberra.

BTRE (2005), Is the world running out of oil? A review of the debate, BTRE Working Paper 61, Canberra.

BTRE (2006a), BTRE Online Transport Statistics. <http://www.btre.gov.au/statistics/statsindex.aspx>

BTRE (2006b), Transport Elasticities Database Online. <http://dynamic.dotars.gov.au/btre/tedb/index.cfm>

Cosgrove, D. (2003), Urban Pollutant Emissions from Motor Vehicles: Australian Trends to 2020, BTRE Final Draft Report for Environment Australia, BTRE. <www.btre.gov.au/docs/joint_reports/urbanpollutants_draft.aspx>

Cosgrove, D. & Mitchell, D. (2001), ‘Standardised Time-series for the Australian Road Transport Task’, Proceedings of the 24th Australasian Transport Research Forum, Hobart, 17th April 2001, Tasmanian Department of Infrastructure, Energy and Resources.

140


Cosgrove, D. & Gargett, D. (1992), ‘The Australian Domestic Transport Task’, Papers of the Australasian Transport Research Forum, vol. 17, part 1, pp. 231–249, BTCE, Canberra.

Cosgrove, D. Evill, B. & Subasic, K. (1994), ‘The Estimation of Transport Emissions—The Social Costs of Emissions’, Papers of the Australasian Transport Research Forum, Vol. 19, Department of Transport, Victoria.

Dynamic Transport Management (1996), Fuel Consumption Project, Final Report, DPIE, Canberra.

Energy Information Administration EIA (2006), Short-Term Energy Outlook, <http://www.eia.doe.gov/emeu/steo/pub/contents.html>

Ensor, M. G. (2002), Methodology to Assess the Benefits of Improved Trip Reliability: Final Report, prepared for Transfund New Zealand by Beca Carter Hollings & Ferner Ltd and Sinclair Knight Merz, August 2002.

FORS (1996), Motor Vehicle Pollution in Australia–Report on the In-Service Vehicle Emissions Study, AGPS, Canberra.

Gargett, D. (2004), ‘Measuring road freight growth’, Papers of the 27th Australasian Transport Research Forum, Adelaide, ATRF.

Gargett, D. & Gafney, J. (2004), ‘Traffic growth in Australian Cities: Causes, Prevention and Cure’, paper for Towards Sustainable Land Transport Conference, 21 November, Wellington, NZ, BTRE staff paper.

Ministry of Transport (1998), Vehicle Fleet Emissions Control Strategy for Local Air Quality Management, Final Report, New Zealand Ministry of Transport, Wellington, NZ.

NGGIC (1998), Workbook for Transport (Mobile Sources), Workbook 3.1 (reprinted Revision 1 with supplements), AGO, Canberra.

NGGIC (2005), Australian Methodology for the Estimation of Greenhouse Gas Emissions and Sinks 2003, Energy (Transport), AGO, Canberra.

NPI (2000), Emissions Estimation Technique Manual for Aggregated Emissions from Motor Vehicles, 22 November 2000–Version 1.0, Aggregated Emissions Data Manual for the National Pollutant Inventory, Environment Australia, Canberra.

141

Further readings

142


Ruimin Li (2004), ‘Examining travel time variability using AVI data’, CAITR paper, December 2004, Institute of Transport Studies, Monash University, Melbourne.

Schipper, L. & Tax, W. (1994), ‘New car test and actual fuel economy: yet another gap?’, Transport Policy, 1994 Vol. 1, No. 4, pg 257-265.

Taylor, M.A.P. & Young, T. (1995), ‘Modelling Emissions and Fuel Consumption for Road Traffic Streams’, paper for WCTR, Sydney 1995.

UK National Atmospheric Emissions Inventory (2004), Database of vehicle emission factors, <www.naei.org.uk/other/vehicle_emissions_v7.xls>.

USEPA (2001), ‘Annex D—Methodology for Estimating Emissions of CH4, N2O, and Criteria Pollutants from Mobile Combustion’.

USEPA (2003), MOBILE6, Technical Documentation Index. <www.epa.gov/otaq/m6.htm>

Waters, M. (1992), Road Vehicle Fuel Economy, Transport and Road Research Laboratory, UK Department of Transport.

Estimating urban traffic and congestion cost trends for ... · of the social costs arising from...

Documents

Transcript of Estimating urban traffic and congestion cost trends for ... · of the social costs arising from...